Web Administrators

15-1299.01

Median wage $116,580/yr435,370 employed (US)Rank #63 of 923 scored · top 7% by substitution

Manage web environment design, deployment, development and maintenance activities. Perform testing and quality assurance of web sites and web applications.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution55

Exposure50

Augmentation80

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

35 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

26%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%50

panel mean rating 3.0/5 → substitution pressure 50/100

Technical feasibility todayw 20%51

panel mean rating 3.0/5 → substitution pressure 51/100

Cost vs. human wagew 15%54

panel mean rating 3.2/5 → substitution pressure 54/100

Adoption barriersw 20%inverted — strong barriers lower the score67

panel mean rating 2.3/5 (barrier strength) → substitution pressure 67/100

Sector adoption velocityw 10%62

panel mean rating 3.5/5 → substitution pressure 62/100

Task breakdown (35 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Track, compile, and analyze Web site usage data.

CI 80–100 · exposure 87 · augmentation 88 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	Web analytics automation is ubiquitous and deeply embedded in digital organizations; nearly all websites use automated tracking, and adoption is mature and fast across all sectors.
Sector adoption velocity	claude-sonnet-5	4/5	Web/IT and digital marketing sectors have rapidly adopted automated analytics and AI-driven reporting tools, with widespread production use of dashboards and automated insights.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered analytics tools augment human analysts by providing real-time dashboards, automated anomaly detection, and intelligent insights, significantly raising their productivity even where human interpretation remains valuable.
Augmentation potential	claude-sonnet-5	5/5	AI tools substantially enhance a web administrator's ability to interpret usage data, surface trends, generate visualizations, and produce narrative summaries, significantly boosting productivity while humans retain oversight.
Task automatability	claude-haiku-4-5-20251001	5/5	Current AI systems can fully automate tracking, compiling, and analyzing website usage data using analytics platforms, log aggregation tools, and statistical analysis. This task is well-defined, data-driven, and involves no subjective judgment, easily meeting the 50% time-saving bar.
Task automatability	claude-sonnet-5	4/5	Web analytics tracking, compilation, and standard reporting is largely automatable with tools like Google Analytics and AI-assisted dashboards that generate insights automatically, though custom interpretation for strategic decisions still needs human input.rating
Adoption barriers	claude-haiku-4-5-20251001	1/5	No licensing, legal, or regulatory barriers prevent automated web analytics; no human sign-off is required, and organizations adopt these tools freely.
Adoption barriers	claude-sonnet-5	1/5	No licensing, liability, or human-contact requirements exist for web analytics tracking and reporting; it's a purely technical/administrative task.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated analytics tools cost a small fraction of hiring a human analyst, often $100–1,000/month for comprehensive tracking versus $60,000+ annual salary, achieving order-of-magnitude cost savings.
Cost vs. human wage	claude-sonnet-5	4/5	Automated analytics platforms with AI reporting cost a small fraction of a human analyst's time for routine data compilation and trend summarization, though oversight for interpretation adds some cost.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Deployed products (Google Analytics, Matomo, Splunk, Datadog, etc.) reliably perform website usage tracking, compilation, and analysis at scale in production across millions of organizations today.
Technical feasibility today	claude-sonnet-5	4/5	Mature production tools (GA4, Adobe Analytics, AI-powered BI dashboards) already automate collection, compilation, and even natural-language summarization of usage data at scale in real organizations.

Document installation or configuration procedures to allow maintenance and repetition.

CI 70–97 · exposure 83 · augmentation 100 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Technology and professional services sectors are rapidly adopting AI writing tools for documentation. Many DevOps and IT teams now use AI assistants in production workflows to generate and maintain runbooks and configuration guides.
Sector adoption velocity	claude-sonnet-5	4/5	IT and software sectors show fast adoption of AI coding/documentation assistants, with widespread integration into developer and sysadmin workflows.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI significantly augments human documentation work by drafting sections, suggesting structure, filling in boilerplate, and auto-generating step lists from logs or code, while the administrator reviews, corrects, and contextualizes the output.
Augmentation potential	claude-sonnet-5	5/5	AI substantially speeds up drafting, formatting, and updating technical documentation while the administrator verifies accuracy and completeness.
Task automatability	claude-haiku-4-5-20251001	5/5	Documenting installation/configuration procedures is largely text generation and transcription of technical steps, which AI excels at. Modern language models can extract procedure details, generate clear documentation, and create repeatable guides with minimal human input, easily meeting the 50% time-saving threshold.
Task automatability	claude-sonnet-5	4/5	AI can generate clear documentation from configuration files, command histories, or screenshots with minimal editing, meeting the time-saving bar for most routine setups.
Adoption barriers	claude-haiku-4-5-20251001	2/5	There are minimal regulatory or legal barriers to AI-generated documentation. The main friction is organizational preference for human review/sign-off and quality oversight, but no licensing requirement mandates human authorship of internal procedures.
Adoption barriers	claude-sonnet-5	1/5	No licensing, regulatory, or liability barriers restrict AI-assisted documentation of technical procedures.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	API costs for LLM-based documentation generation are negligible (often cents per document) compared to the loaded hourly wage of a web administrator (typically $50–100+/hour) for the same output, yielding an order-of-magnitude cost advantage.
Cost vs. human wage	claude-sonnet-5	4/5	Generating documentation via LLM inference is far cheaper than the time an admin spends writing detailed procedural docs manually, though some review overhead remains.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Multiple deployed products (GitHub Copilot, ChatGPT, Notion AI, specialized documentation tools) reliably generate technical documentation at scale in production. Organizations actively use AI to auto-generate and maintain installation guides with high consistency.
Technical feasibility today	claude-sonnet-5	3/5	Tools like Copilot, ChatGPT, and internal doc-generation bots are used in production to draft runbooks and READMEs, but accuracy on complex or environment-specific configs still requires human review.

Check and analyze operating system or application log files regularly to verify proper system performance.

CI 75–86 · exposure 80 · augmentation 88 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	IT and web hosting sectors have deeply and widely adopted automated log monitoring and observability platforms; this is standard practice in most digitized organizations and cloud-native environments, not a pilot-stage experiment.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and web infrastructure sectors have rapidly adopted AI-driven observability and monitoring tools, with automated log analysis now standard practice in many production environments.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI log analysis tools substantially augment human operators by surfacing anomalies, correlating events, and ranking alerts by severity, allowing administrators to focus on root cause and remediation rather than manual log scanning. This transforms productivity while the human remains in the loop for decision-making.
Augmentation potential	claude-sonnet-5	5/5	AI-based log analysis and anomaly detection significantly enhance an administrator's ability to spot issues quickly, prioritize alerts, and reduce noise, while the human remains responsible for interpretation and action.
Task automatability	claude-haiku-4-5-20251001	4/5	Log file analysis is highly automatable via pattern matching, anomaly detection, and rule-based systems. Current AI tools (e.g., ELK Stack, Splunk, ML-driven monitoring) can parse, categorize, and flag issues in real-time, easily achieving 50% time savings. Full end-to-end automation is limited only by the occasional need for human judgment on novel system states, not fundamental technical barriers.
Task automatability	claude-sonnet-5	4/5	Log parsing, anomaly detection, and summarization of performance issues are well-suited to automated pattern matching and LLM-based analysis, saving substantial manual review time.But full root-cause diagnosis and remediation decisions still often need human judgment, so it's not fully end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few regulatory or licensing barriers exist for automating log analysis; it is an informational task with no required human sign-off or legal gatekeeping. The main friction is organizational inertia and the need for tuning alerting thresholds, but these do not meaningfully prevent automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing or legal requirement mandates human review of logs; the main friction is organizational trust in automated alerts and the need for human oversight on critical infrastructure decisions.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated log analysis via cloud monitoring platforms costs a fraction of a human FTE per month, even accounting for overhead and false-positive review. The cost per analysis cycle is at least an order of magnitude cheaper than paying a web administrator to manually review logs on a regular schedule.
Cost vs. human wage	claude-sonnet-5	4/5	Automated log analysis tools process massive volumes of log data far cheaper than manual review by an administrator, though licensing and infrastructure costs for enterprise-grade platforms keep it from being a full order-of-magnitude cheaper in all cases.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Multiple mature, production-deployed systems reliably perform log analysis at scale today (Splunk, Datadog, New Relic, ELK Stack, CloudWatch Logs). These platforms are used across thousands of organizations for automated log monitoring, alerting, and anomaly detection with well-established accuracy and operational integration.
Technical feasibility today	claude-sonnet-5	4/5	Mature log-monitoring products (Splunk, Datadog, ELK with ML-based anomaly detection, AI-assisted observability tools) are widely deployed in production and reliably flag anomalies and performance issues today.

Inform Web site users of problems, problem resolutions, or application changes and updates.

CI 72–84 · exposure 80 · augmentation 100 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Tech companies and digital organizations have rapidly adopted automated status pages and incident notification systems; production deployment is common in DevOps and SRE teams.
Sector adoption velocity	claude-sonnet-5	3/5	IT/web administration is a digitized, tech-forward sector with moderate AI tool adoption for communications, though full automation of user notifications is still emerging rather than universal.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI significantly assists humans by auto-drafting notifications, suggesting severity levels, and flagging stakeholders, allowing administrators to focus on problem-solving rather than communication logistics.
Augmentation potential	claude-sonnet-5	5/5	AI is highly effective at drafting, summarizing, and tailoring problem/update notifications for administrators to quickly review and send, significantly speeding this communication task.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can generate, draft, and distribute notifications about outages, fixes, and updates with minimal human involvement; tools can monitor systems, compose status messages, and push them via email/dashboards. Remaining 20–30% requires human judgment on tone, severity categorization, and stakeholder-specific messaging.
Task automatability	claude-sonnet-5	4/5	Drafting and sending status updates, incident notices, and changelog communications is largely templated text generation that current LLMs handle well end-to-end with human review.'
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few regulatory or licensing barriers exist; oversight is light (mostly organizational preference for human review). Some organizations may prefer human sign-off on external communications, but no hard legal requirement blocks automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizations may want human review to avoid miscommunication about outages or security issues, creating light oversight friction.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated monitoring and AI-assisted notification generation cost pennies per message, orders of magnitude cheaper than manual composition, review, and distribution by a human administrator.
Cost vs. human wage	claude-sonnet-5	4/5	Generating and distributing routine notifications via AI is far cheaper than having a human admin manually compose each update, though some oversight cost remains.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Deployed monitoring and notification platforms (PagerDuty, Statuspage, Datadog) routinely auto-generate and send incident notifications; LLMs can draft and customize status updates in production systems today.
Technical feasibility today	claude-sonnet-5	4/5	Status page tools and AI-drafted incident communications (e.g., automated status.io/Statuspage-style updates, AI-generated release notes) are already deployed in production at many organizations.

Review or update Web page content or links in a timely manner, using appropriate tools.

CI 70–84 · exposure 70 · augmentation 88 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Web-focused organizations (tech, media, SaaS, e-commerce) widely deploy automated link checkers and content management integrations. Adoption is fast in digital-native sectors, though smaller or less digitalized organizations may lag.
Sector adoption velocity	claude-sonnet-5	4/5	Web/IT and digital marketing sectors have rapidly adopted automation tools (CMS bots, SEO/link-checking software, AI content assistants) as standard practice.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI tools substantially assist administrators by identifying broken links, suggesting outdated content, and bulk-updating metadata. The human retains final approval, but AI dramatically accelerates the review and flagging workflow.
Augmentation potential	claude-sonnet-5	5/5	AI tools substantially boost productivity for administrators by flagging broken links, suggesting content updates, and automating routine checks while humans retain final approval.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI can identify outdated links, suggest content updates, and automate routine link checking and broken-link fixes using standard tools. However, evaluating timeliness requirements and deciding which content truly needs updating still requires some human judgment, preventing a full 5 rating.
Task automatability	claude-sonnet-5	4/5	Content updates, link checking, and broken-link remediation are largely rule-based and text-driven, tasks current AI/agent tools can handle with substantial time savings, though some judgment on content relevance remains.
Adoption barriers	claude-haiku-4-5-20251001	2/5	No licensing or regulatory requirement mandates human review of web pages. Most organizations retain oversight for brand/accuracy reasons, but no hard legal barrier prevents full automation of link and content updates.
Adoption barriers	claude-sonnet-5	1/5	No licensing, regulatory, or liability barriers restrict automating routine web content and link maintenance.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated link checking and content scanning cost fractions of a cent per page; a human web administrator reviewing pages manually costs $25–50+ per hour. The cost differential is easily an order of magnitude in favor of AI automation.
Cost vs. human wage	claude-sonnet-5	4/5	Automated crawlers, link-checking bots, and AI-assisted content editors are cheap to run continuously compared to paying a web administrator's hourly rate for the same routine checks.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (link checkers, content management systems with AI suggestions, automated SEO tools) reliably perform link review and flagging of stale content. Some manual verification of content accuracy remains standard practice, but the core technical work is mature and in production.
Technical feasibility today	claude-sonnet-5	3/5	CMS platforms with AI plugins, link-checkers, and automated content pipelines exist and are used in production, but fully autonomous review/update without human oversight is not yet standard for most organizations.

Document application and Web site changes or change procedures.

CI 70–75 · exposure 70 · augmentation 88 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Information technology and software development sectors—where this task is concentrated—are rapidly adopting AI-assisted documentation tools, with many teams already using automated changelog and README generation as standard practice.
Sector adoption velocity	claude-sonnet-5	4/5	IT and software development sectors show fast adoption of AI-assisted documentation and DevOps tooling, with many teams integrating AI into changelogs and commit summaries already.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially augments human documentation work by auto-generating drafts from logs and diffs, freeing engineers to focus on narrative clarity, policy implications, and cross-referencing, significantly raising productivity in this routine task.
Augmentation potential	claude-sonnet-5	5/5	AI tools substantially speed up drafting of technical documentation and change logs while humans verify and finalize, representing a strong augmentation use case.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can substantially automate documentation of changes by analyzing code diffs, commit logs, and deployment records to generate structured change summaries; however, capturing context-specific business rationale and compliance nuances typically requires human review, preventing full end-to-end automation at equal quality.
Task automatability	claude-sonnet-5	4/5	Generating documentation from code diffs, commit logs, and configuration changes is a well-structured text generation task that LLMs handle well, especially with tool access to version control systems., allowing significant time savings.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Documentation is typically a best-practice requirement rather than a legally mandated task, and organizations face minimal regulatory or liability barriers to automating it; the main friction is ensuring coverage completeness and quality control.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but some organizations impose internal review/audit standards for change documentation, creating mild friction rather than a hard barrier.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven documentation generation costs a fraction of manual human time (typically $0.01–$0.10 per document versus $50–$200 in loaded labor), yielding substantial savings when integrated into CI/CD pipelines.
Cost vs. human wage	claude-sonnet-5	4/5	Automated changelog and documentation generation via LLMs is very cheap compared to a web administrator's time spent manually writing documentation.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (GitHub Copilot, documentation-generation tools, LLM APIs) reliably extract and summarize technical changes from version control and logs in production environments, though they still benefit from human verification for completeness and accuracy.
Technical feasibility today	claude-sonnet-5	3/5	AI coding assistants and documentation tools (e.g., GitHub Copilot, changelog generators) are deployed and used for this today, but still require human review to ensure accuracy and completeness for compliance/audit purposes.

Implement updates, upgrades, and patches in a timely manner to limit loss of service.

CI 46–95 · exposure 67 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	Cloud-native and DevOps-heavy sectors (SaaS, finance, tech) have deeply adopted automated patching for years; even traditional enterprises increasingly deploy CI/CD. Public benchmarks show this is among the most-adopted IT automation patterns.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations broadly uses automation tools (Ansible, patch management suites) with growing AI integration, but fully autonomous patch decision-making is still uncommon in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI systems assist humans by suggesting patch schedules, predicting deployment risks, monitoring rollback triggers, and alerting on failures. The human remains in oversight, but productivity gains are substantial compared to purely manual workflows.
Augmentation potential	claude-sonnet-5	4/5	AI-assisted monitoring, vulnerability scanning, and automated patch scheduling significantly boost administrator efficiency while humans retain oversight for critical decisions.
Task automatability	claude-haiku-4-5-20251001	5/5	This task is highly automatable using modern CI/CD pipelines, configuration management tools (Ansible, Terraform), and cloud platform auto-scaling. Current systems can detect updates, stage patches, run tests, and deploy with minimal human intervention, easily achieving >50% time savings at equal or better quality.
Task automatability	claude-sonnet-5	2/5	Applying patches involves judgment about compatibility, timing, rollback planning, and testing that current AI cannot fully own end-to-end without human oversight, though scripting and scheduling parts can be automated.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While some organizations require manual approval gates for compliance or business continuity reasons, no legal or licensing barrier prevents automated deployment. Most friction is organizational risk tolerance rather than regulatory constraint.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational risk aversion around service outages and change-management approval processes create moderate friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Infrastructure automation tools cost far less than full-time human operators managing patches manually; a single automation pipeline handles thousands of systems for pennies per transaction versus hours of human labor.
Cost vs. human wage	claude-sonnet-5	3/5	Automated patch deployment tools reduce labor costs substantially, but human validation, monitoring, and incident response for failures keep overall cost comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Deployed products (GitHub Actions, GitLab CI, AWS Systems Manager, Kubernetes operators) reliably handle patch deployment and upgrades in production at scale across millions of organizations. This is standard infrastructure automation in 2024.
Technical feasibility today	claude-sonnet-5	3/5	Automated patch management tools and CI/CD pipelines with AI-assisted testing exist in production, but full autonomous patching with risk assessment across heterogeneous systems remains narrow and error-prone.

Set up or maintain monitoring tools on Web servers or Web sites.

CI 57–84 · exposure 67 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	IT and web operations sectors show rapid, deep adoption of infrastructure automation and managed monitoring services. DevOps practices are standard in tech, finance, and high-digitization industries; adoption is measured and production-ready rather than experimental.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and DevOps sectors are rapidly adopting AI-assisted tooling (AIOps, copilot-based config generation) with notable production deployment in tech-forward organizations.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-assisted monitoring configuration and anomaly detection significantly amplify a web administrator's productivity by automating alert tuning, suggesting optimal metric collection, and accelerating troubleshooting workflows while keeping the human in control of critical infrastructure decisions.
Augmentation potential	claude-sonnet-5	4/5	AI significantly speeds up writing monitoring scripts, generating alert rules, and troubleshooting configurations, while humans still validate and adapt to their specific infrastructure.
Task automatability	claude-haiku-4-5-20251001	4/5	Setting up and maintaining monitoring tools on web servers is highly repetitive and scriptable. Current CI/CD and infrastructure-as-code tools can automate most deployment, configuration, and routine maintenance tasks; however, diagnosing novel system failures or tuning alerts for edge cases still typically requires human judgment, keeping it slightly below full automation.
Task automatability	claude-sonnet-5	3/5	AI can generate monitoring configurations, scripts, and dashboards (e.g., Prometheus, Grafana, uptime checks) but selecting appropriate metrics, thresholds, and integrating with existing infrastructure still requires human judgment and validation.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few regulatory or licensing barriers exist for automating monitoring tool setup and maintenance; however, organizational friction around change management, security audit requirements, and the preference to maintain human oversight of infrastructure changes create modest adoption friction.
Adoption barriers	claude-sonnet-5	2/5	No licensing or regulatory requirement mandates a human perform this task, though organizations may require review of monitoring configs due to security and uptime risk.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI-driven infrastructure automation and monitoring tools cost a fraction of the fully-loaded wage of a web administrator performing these repetitive setup and maintenance tasks. The marginal cost of additional monitoring deployments via automation is near-zero at scale.
Cost vs. human wage	claude-sonnet-5	3/5	AI tools reduce setup time and lower costs versus a human doing it manually, but ongoing maintenance, tuning, and incident response still require paid human oversight, keeping costs roughly comparable to partial automation savings.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature products like Prometheus, DataDog, New Relic, and Grafana are widely deployed in production at scale. Infrastructure automation tools (Terraform, Ansible) and monitoring-as-a-service platforms handle setup and maintenance reliably across thousands of organizations.
Technical feasibility today	claude-sonnet-5	3/5	AI coding assistants and DevOps copilots can scaffold monitoring setups and suggest alerting rules in production use, but reliable end-to-end autonomous setup across diverse server environments is not yet standard practice.

Develop Web site performance metrics.

CI 57–84 · exposure 62 · augmentation 88 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Web hosting and DevOps sectors have rapidly adopted automated monitoring solutions; most organizations already run APM platforms, indicating fast, deep adoption in digitized environments.
Sector adoption velocity	claude-sonnet-5	4/5	Web administration and IT/analytics functions are in a fast-adopting sector where AI-driven monitoring and metrics tools are already widely integrated into DevOps and web operations pipelines.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI and automation significantly assist humans by instantly surfacing trends, anomalies, and correlations that would take hours to spot manually, enabling faster decision-making while the admin interprets and acts on insights.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially help by suggesting relevant metrics, benchmarking against industry standards, and automating dashboard creation, significantly boosting a human administrator's productivity in this task.
Task automatability	claude-haiku-4-5-20251001	4/5	Most of the work—defining metrics, setting up monitoring dashboards, collecting performance data, and generating reports—can be automated with current tools (APM platforms, analytics libraries, CI/CD integrations). However, deciding which metrics align with business goals typically requires human judgment, leaving ~20–30% of the task requiring oversight.
Task automatability	claude-sonnet-5	3/5	AI can help draft metric frameworks (e.g., Core Web Vitals, uptime, latency) and even generate monitoring configs, but selecting business-relevant KPIs and thresholds still requires human judgment about organizational goals and context, so only part of the task is automatable end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few legal or regulatory barriers exist for automated metric collection and reporting; organizations may prefer human validation of metrics strategy, but nothing prevents technical automation of the execution itself.
Adoption barriers	claude-sonnet-5	1/5	There are no licensing, legal, or human-contact requirements around defining web performance metrics; it's a purely technical/organizational task with no regulatory barrier.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Cloud-based APM and analytics platforms cost a fraction of a web admin's loaded wage per metric set, with marginal costs near zero for additional metrics once the platform is in place.
Cost vs. human wage	claude-sonnet-5	3/5	Using AI tools to draft or configure metrics is cheap relative to a dedicated analyst's time, but integration with existing systems and validation still requires paid human oversight, keeping costs roughly comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature products (Datadog, New Relic, Splunk, Google Analytics) reliably perform metric definition, collection, and dashboard creation in production at scale. Integration is straightforward; the main gap is that humans still select what to measure.
Technical feasibility today	claude-sonnet-5	3/5	Analytics and monitoring products (e.g., Google Analytics, Datadog, New Relic) already include AI-assisted metric suggestions and anomaly detection, but defining bespoke performance metrics for a specific site's goals is not fully productized as an autonomous capability.

Correct testing-identified problems, or recommend actions for their resolution.

CI 55–84 · exposure 70 · augmentation 88 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Tech and SaaS companies are rapidly adopting AI-powered testing and remediation agents in production DevOps pipelines. Continuous integration platforms routinely use LLMs for code review and bug fixing, representing deep sector-wide adoption momentum.
Sector adoption velocity	claude-sonnet-5	3/5	placeholder
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI dramatically amplifies web administrator productivity by instantly surfacing root causes, generating fix suggestions, and automating routine remediation while the human reviews and approves changes. This assistive role is transformative for problem diagnosis and solution velocity.
Augmentation potential	claude-sonnet-5	4/5	placeholder
Task automatability	claude-haiku-4-5-20251001	5/5	AI systems can now autonomously identify, diagnose, and recommend fixes for web testing failures (broken links, performance issues, security vulnerabilities) with substantial time savings. Modern testing frameworks integrated with LLMs can generate corrective code and mitigation strategies, meeting the ≥50% time-saving threshold.
Task automatability	claude-sonnet-5	3/5	AI coding assistants can diagnose common bugs, config errors, and suggest fixes from logs/test output, but complex infrastructure or ambiguous issues still need human judgment and system context.5
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few regulatory or legal barriers prevent AI-assisted or autonomous testing-problem resolution in most jurisdictions. Most organizations rely on automated CI/CD pipelines with minimal human sign-off, and AI recommendations can integrate seamlessly into existing workflows.
Adoption barriers	claude-sonnet-5	2/5	placeholder
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven code analysis and recommendation systems cost far less than hiring skilled web administrators for routine issue triage and remediation. Inference overhead is minimal compared to human diagnostic labor, yielding significant cost advantages on standard problem classes.
Cost vs. human wage	claude-sonnet-5	3/5	placeholder
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products like GitHub Copilot, ChatGPT for code analysis, and specialized testing tools reliably identify and suggest fixes for common web testing issues in production pipelines. Error rates on straightforward problems are low, though complex architectural issues may require human expertise.
Technical feasibility today	claude-sonnet-5	3/5	placeholder

Monitor systems for intrusions or denial of service attacks, and report security breaches to appropriate personnel.

CI 53–81 · exposure 67 · augmentation 100 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	Security monitoring automation is deeply embedded in enterprise IT and financial-services operations; SIEM and IDS adoption is near-universal among organizations handling sensitive data, with rapid deployment of newer AI-driven threat detection in high-digitization sectors.
Sector adoption velocity	claude-sonnet-5	4/5	Cybersecurity and IT operations are among the faster-adopting fields, with AI-based threat detection tools now standard in mid-to-large organizations' security stacks.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI-driven security tools dramatically augment human analyst productivity by filtering noise, prioritizing alerts, and correlating events across systems, allowing analysts to focus investigation and response efforts on genuine high-severity threats while remaining in the decision loop.
Augmentation potential	claude-sonnet-5	5/5	AI significantly enhances human capability in this domain by continuously monitoring high-volume traffic, correlating signals, and surfacing likely threats, letting analysts focus on validated incidents while remaining in the loop for reporting.
Task automatability	claude-haiku-4-5-20251001	4/5	Security monitoring tools can autonomously detect intrusions and DoS attacks using pattern matching and anomaly detection, then generate alerts and logs for human review. The task of continuous surveillance and threat detection is highly automatable, though final breach classification and escalation decisions may benefit from human oversight.
Task automatability	claude-sonnet-5	3/5	AI-driven monitoring tools can detect anomalies and flag potential intrusions automatically, but final judgment, escalation decisions, and nuanced investigation still require human oversight, so it's roughly half automatable end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Regulatory requirements (SOC 2, HIPAA, PCI-DSS) and compliance frameworks often mandate human verification and sign-off on security breaches, and organizational liability concerns create friction around fully autonomous response without human oversight, though monitoring automation itself faces few legal barriers.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement for this specific task, but organizational risk tolerance, liability for missed breaches, and compliance frameworks (e.g., SOC2, HIPAA) create moderate friction around fully automating breach reporting decisions.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated SIEM and monitoring platforms typically cost far less per threat-instance detected than the loaded cost of human security analysts conducting 24/7 network surveillance, often delivering an order-of-magnitude cost advantage when amortized across organizational scale.
Cost vs. human wage	claude-sonnet-5	3/5	Automated monitoring tools reduce headcount needs but still require licensing costs, integration, and human security analysts for interpretation and response, making costs roughly comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature security information and event management (SIEM) systems, intrusion detection systems (IDS), and DoS mitigation platforms are deployed at scale in production environments today, reliably detecting and alerting on security threats with well-established benchmarks and vendor ecosystems.
Technical feasibility today	claude-sonnet-5	3/5	Deployed SIEM/IDS products with ML-based anomaly detection (e.g., Splunk, CrowdStrike, Darktrace) are widely used in production, but false positive/negative rates remain material and human analysts still triage alerts.

Back up or modify applications and related data to provide for disaster recovery.

CI 53–81 · exposure 62 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	5/5	Disaster recovery automation and backup orchestration are core practices in modern DevOps and cloud operations. Large-scale adoption is already established in information technology, financial services, and enterprise environments, with automation commonplace in production.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and cloud infrastructure sectors have rapidly adopted automation and AIOps tooling for backup, monitoring, and infrastructure management, though full DR automation remains an active but growing area.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI tools significantly augment web administrators by automating routine backup and recovery configuration tasks while administrators focus on disaster recovery strategy, testing protocols, and handling exceptional scenarios that require human judgment.
Augmentation potential	claude-sonnet-5	4/5	AI and automation scripts significantly boost efficiency in scheduling backups, detecting anomalies, and orchestrating recovery steps, meaningfully augmenting administrators while they retain oversight of critical recovery decisions.
Task automatability	claude-haiku-4-5-20251001	4/5	The task of backing up and modifying applications/data for disaster recovery is highly structured and repetitive, with established protocols and tooling. Current AI systems can orchestrate backup workflows, trigger data snapshots, modify configurations, and verify recovery procedures with minimal human intervention, achieving well over 50% time savings at equal quality.
Task automatability	claude-sonnet-5	3/5	Backup automation and scripted disaster recovery workflows are well-established, but modifying applications for recovery scenarios and validating recovery integrity still require human judgment and oversight, so only part of the task meets the 50% time-saving bar.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While some organizations impose change-control policies and require human sign-off on production modifications, there are no legal or licensing barriers preventing automated backup and recovery workflows. Most friction is organizational process rather than hard regulatory requirement.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement, but organizational risk tolerance is low given the high cost of DR failures, creating oversight and validation requirements that slow full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated backup and disaster recovery solutions cost a fraction of manual labor once deployed. Infrastructure-as-code and cloud-native backup services operate at near-zero marginal cost per execution compared to a web administrator's loaded hourly wage.
Cost vs. human wage	claude-sonnet-5	3/5	Automated backup tooling is cheap to run at scale, but designing, testing, and maintaining DR plans and validating recovery still requires skilled engineer time, keeping overall cost roughly comparable to a human-managed process augmented by tools.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature DevOps and infrastructure automation tools (Terraform, Ansible, CloudFormation, backup orchestration platforms) reliably perform backup and recovery modifications in production environments at scale. While some complex edge cases may require human oversight, the core task is deployable and widely in use.
Technical feasibility today	claude-sonnet-5	3/5	Mature backup/DR products (cloud snapshotting, automated failover, IaC tools) exist and are widely deployed, but AI-driven autonomous management of full backup/recovery pipelines with application-level modification is not yet standard practice.

Install or configure Web server software or hardware to ensure that directory structure is well-defined, logical, and secure, and that files are named properly.

CI 54–75 · exposure 62 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Cloud-native and DevOps practices have achieved rapid, deep adoption in tech, finance, and large enterprises. Infrastructure automation is a standard practice in digitized sectors, though smaller organizations and legacy-heavy industries lag considerably.
Sector adoption velocity	claude-sonnet-5	3/5	IT/software sectors are moderately fast adopters of AI-assisted DevOps tooling, though full automation of server administration is still pilot-stage in many organizations.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered configuration assistants and security scanners substantially improve a web administrator's productivity by generating templates, validating configurations, and identifying security issues. The human remains in control but accomplishes far more in the same time.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially speed up drafting configuration files, directory schemas, and naming conventions, letting administrators focus on validation and security review.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI systems can handle most directory structure planning, naming conventions, and security configuration through infrastructure-as-code generation and deployment automation. However, context-specific decisions about organizational needs and legacy system compatibility still typically require human oversight, preventing a full 5 rating.
Task automatability	claude-sonnet-5	3/5	AI coding agents can generate server configs, directory structures, and naming conventions, but real deployments require environment-specific decisions, hardware interaction, and validation that still need human oversight for a majority of production cases.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Organizations often require human sign-off on security configurations and directory structures due to compliance, liability, and audit trails. Regulatory requirements (especially in finance, healthcare) and the need for documented accountability create moderate friction against full automation without human approval.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but production systems often require security review and organizational change-control processes before configurations go live, creating moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	Automation tooling and cloud-based deployment services cost substantially less than manual server configuration by salaried administrators once amortized across deployments. A single configuration template can serve hundreds of instances, creating significant cost advantage over human labor.
Cost vs. human wage	claude-sonnet-5	4/5	Once configured, AI-assisted scripting and templating dramatically cuts time versus manual setup, though initial integration and validation by a skilled admin still adds cost.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature DevOps and Infrastructure-as-Code tools (Terraform, Ansible, configuration management systems) reliably automate server configuration in production environments. Some gaps remain in real-time troubleshooting and novel security scenarios, but core provisioning is demonstrably deployed at scale.
Technical feasibility today	claude-sonnet-5	3/5	DevOps copilots and IaC tools (Terraform, Ansible with AI assistance) can scaffold and configure web servers, but reliable end-to-end autonomous configuration in production environments is still narrow and error-prone for complex or legacy setups.

Develop or document style guidelines for Web site content.

CI 41–84 · exposure 55 · augmentation 88 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Web administration is reasonably digitized and early-adopter sectors use AI for documentation assistance, but actual production adoption of fully AI-generated style guidelines remains limited because organizations view these as strategic, brand-critical decisions requiring human ownership.
Sector adoption velocity	claude-sonnet-5	4/5	Web administration and content operations sit within tech/digital sectors that have adopted generative AI writing tools quickly and at meaningful depth for documentation tasks.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists web administrators by generating initial drafts, providing templates, suggesting best practices, and automating documentation formatting, allowing humans to focus on customization, review, and strategic decisions rather than starting from scratch.
Augmentation potential	claude-sonnet-5	5/5	AI is highly effective as a drafting and brainstorming aid for style guides, letting humans quickly iterate and refine language, structure, and examples while retaining final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can generate initial style guidelines drafts and documentation templates with reasonable quality, but developing comprehensive, organization-specific guidelines requires understanding brand identity, user needs, and strategic goals that need human judgment. Significant human review and refinement would be needed, so time savings fall short of 50% with equal quality.
Task automatability	claude-sonnet-5	4/5	Drafting style guides is largely a language/document synthesis task, which current LLMs handle well by producing structured, comprehensive guidelines from prompts or examples with human review saving significant time.astro.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While there are no strict legal requirements that a licensed human must develop style guidelines, most organizations prefer human accountability for brand consistency and strategic decisions. Documentation and style guide creation often falls within broader web administrator responsibilities with organizational oversight expectations.
Adoption barriers	claude-sonnet-5	1/5	There is no licensing, legal, or regulatory requirement that a human author style guidelines, and no liability concerns attach to this internal documentation task.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI inference costs for generating style guideline drafts are low, but the task requires significant human oversight, testing, and revision to ensure quality and organizational fit, making the total cost comparable to having a human web administrator develop guidelines from scratch.
Cost vs. human wage	claude-sonnet-5	5/5	Generating a draft style guide via an LLM costs a few cents to dollars in compute versus hours of a web administrator's paid time, making AI drastically cheaper for this specific output.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products like Claude, ChatGPT, and specialized documentation tools can draft style guidelines and content standards, but real-world deployment shows these outputs often require substantial customization and human expertise to ensure alignment with organizational brand voice and technical requirements.
Technical feasibility today	claude-sonnet-5	4/5	Products like ChatGPT, Claude, and specialized content-ops tools are routinely used in production to draft and maintain style guides, though customization for brand voice still requires human editing.

Gather, analyze, or document user feedback to locate or resolve sources of problems.

CI 57–62 · exposure 50 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Web administration and IT operations are digitized sectors with active adoption of AI-driven monitoring, ticketing automation, and chatbots. Production deployments of feedback analysis tools are increasingly common in tech-forward organizations.
Sector adoption velocity	claude-sonnet-5	4/5	IT and web operations roles are within fast-adopting tech/professional services sectors, with sentiment analysis and support-ticket automation already common.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI meaningfully assists by automatically summarizing feedback, highlighting patterns, and prioritizing high-frequency issues, allowing administrators to focus investigative effort where it matters most while human judgment remains central to diagnosis.
Augmentation potential	claude-sonnet-5	4/5	AI significantly speeds up gathering, categorizing, and summarizing feedback, letting administrators focus on prioritization and actual fixes.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can automate parts of this task—gathering feedback from support tickets, emails, and surveys; performing sentiment analysis; clustering issues by theme—but resolving root causes and understanding context often requires human judgment and domain knowledge. Current systems handle data collection and pattern detection but struggle with nuanced problem diagnosis.
Task automatability	claude-sonnet-5	3/5	AI can aggregate and summarize user feedback and cluster common issues, but diagnosing actual root causes often requires cross-referencing logs, code, and system context that still needs human investigation.atn
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few legal or licensing barriers exist; organizations can adopt AI-assisted feedback tools freely. Primary friction comes from preferences for human judgment on critical issues and the need for human sign-off on root-cause findings.
Adoption barriers	claude-sonnet-5	1/5	There are no licensing or regulatory requirements around analyzing user feedback for a website; organizational tooling is the main constraint.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI inference and integration costs for feedback analysis are modest, but oversight and validation by skilled web administrators remain necessary, making the all-in cost competitive with rather than dramatically cheaper than human effort alone.
Cost vs. human wage	claude-sonnet-5	3/5	AI tools reduce time spent on triage and summarization, but human oversight and validation of findings still add cost, keeping the overall ratio moderate rather than dramatically cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Multiple deployed products (ticketing systems with AI triage, chatbots for feedback collection, sentiment analysis tools) perform components reliably, but end-to-end problem resolution requires human oversight. Error rates remain material on complex or novel issues.
Technical feasibility today	claude-sonnet-5	3/5	Products like AI-powered ticketing systems, sentiment analyzers, and log analysis tools exist and are used in production, but reliable end-to-end root-cause identification from feedback alone remains narrow and error-prone.

Provide training or technical assistance in Web site implementation or use.

CI 41–75 · exposure 50 · augmentation 88 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Information and technology sectors (where web administration is concentrated) are rapidly integrating AI chatbots and automation into help desk and training delivery; many firms already use AI-assisted knowledge bases and automated support tiers.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web development sectors show moderate AI tool adoption (e.g., AI coding/documentation assistants), but structured training/technical assistance workflows are still often human-led with AI as a pilot-stage aid.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI excels at augmenting web administrators through real-time documentation lookup, code snippet generation, documentation drafting, and instant troubleshooting suggestions, materially raising their ability to assist multiple users concurrently while they retain oversight.
Augmentation potential	claude-sonnet-5	4/5	AI significantly aids by generating documentation, answering common technical questions, and drafting training materials, letting human trainers focus on nuanced or complex issues.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can draft training materials, answer FAQ-style technical questions, troubleshoot common issues, and generate documentation with minimal human oversight, saving >50% of time on routine implementation guidance. Complex bespoke training still requires human judgment but represents a minority of effort.
Task automatability	claude-sonnet-5	2/5	Training and technical assistance require live interaction, adapting to specific user needs and context, which current AI can support but not fully replace end-to-end at equal quality without significant human oversight.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Training and assistance have modest barriers—no licensing requirement, liability is limited (incorrect advice causes rework not safety risk), and while organizations may prefer human support for complex implementations, nothing legally mandates it.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational trust, need for tailored guidance, and liability for misconfigurations create moderate friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	API inference costs for generating training responses are typically <$1 per user interaction versus $50–150/hour loaded wage for technical staff, yielding 50–100x cost advantage even with oversight overhead.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted documentation and support chatbots can reduce costs for routine questions, but complex or customized training still requires human involvement, keeping overall cost roughly comparable to skilled labor.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (ChatGPT, Claude, specialized support bots) demonstrate reliable performance on typical web administration Q&A, documentation, and basic troubleshooting in production systems, though edge cases and novel architectures sometimes require human correction.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI documentation tools exist for basic help, but delivering reliable, contextualized training or troubleshooting for specific web implementations is not yet a mature deployed product capability.

Identify or document backup or recovery plans.

CI 43–67 · exposure 53 · augmentation 88 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI for backup and recovery planning is limited; most organizations treat these tasks as high-stakes and prefer human expertise, with only early-stage pilots in tech-forward firms using AI assistants.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web administration functions are moderately fast adopters of AI tools for documentation, but formal DR planning workflows are often still manual and pilot-stage.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments this task by generating plan templates, identifying gaps in current documentation, analyzing system configurations for recovery points, and automating routine updates while the administrator maintains oversight and final validation.
Augmentation potential	claude-sonnet-5	5/5	AI significantly speeds up drafting, structuring, and updating recovery documentation while administrators remain responsible for verifying technical accuracy and completeness.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can assist in identifying and documenting portions of backup and recovery plans by analyzing existing infrastructure and generating template documentation, but requires significant human oversight to ensure plans are complete, aligned with organizational risks, and compliant with specific business requirements.
Task automatability	claude-sonnet-5	4/5	Drafting backup/recovery plans is largely a documentation and standardization task that LLMs can produce from system inventories and best-practice templates, requiring human review for accuracy and completeness.'
Adoption barriers	claude-haiku-4-5-20251001	3/5	Documentation automation faces moderate friction from organizational requirements for human sign-off, verification of disaster recovery procedures, and liability concerns over plan completeness and accuracy; some regulatory contexts may require human certification.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement exists for writing backup plans, though organizational sign-off and accuracy verification create some friction before AI-drafted plans are trusted.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI-assisted documentation can reduce time spent on routine plan creation and updates, making costs roughly comparable to human work when accounting for quality assurance and customization overhead.
Cost vs. human wage	claude-sonnet-5	4/5	Generating and updating documentation via AI is much cheaper than a human spending hours drafting from scratch, though some cost remains for review and validation.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI tools can help with documentation generation and some analysis, no mature product reliably identifies comprehensive backup and recovery plans end-to-end; most require substantial human expertise to validate coverage and test viability.
Technical feasibility today	claude-sonnet-5	3/5	AI writing assistants and IT documentation tools can generate draft disaster recovery/backup plans today, but they still require significant human verification against actual infrastructure, so production reliability is moderate.

Develop testing routines and procedures.

CI 51–57 · exposure 50 · augmentation 75 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Tech and software sectors show moderate adoption of AI-assisted testing (copilots, generative test tools in CI/CD pipelines), with pilots widespread but full production deployment still emerging. Early-stage adoption patterns rather than deep market penetration.
Sector adoption velocity	claude-sonnet-5	4/5	Software development and IT sectors are among the fastest adopters of AI coding and testing tools, with widespread use of AI-assisted development already common practice.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at augmenting test designers by rapidly generating candidate test cases, covering edge cases, and maintaining test suites as code changes. Human testers benefit substantially from AI suggestions while retaining control over coverage strategy and acceptance criteria.
Augmentation potential	claude-sonnet-5	4/5	AI significantly speeds up drafting of test cases, edge-case identification, and procedure documentation, meaningfully boosting administrator productivity while they retain oversight of final testing strategy.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can generate test cases, automate test script creation, and suggest testing strategies, cutting manual test design time substantially. However, defining business-critical test coverage and acceptance criteria typically requires domain expertise and human judgment, so full end-to-end automation remains incomplete.
Task automatability	claude-sonnet-5	3/5	AI can draft test cases, scripts, and QA procedures from specifications, but developing comprehensive testing routines for a specific web infrastructure requires contextual knowledge and validation that limits full automation today.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Testing procedures do not require legal licensing or mandatory human sign-off, and AI tools integrate readily into development pipelines. Light barriers exist around organizational adoption culture and QA team buy-in, but nothing prevents technical substitution.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement exists for this task, though organizational review and quality assurance processes create moderate friction before AI-generated testing procedures are trusted in production.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-assisted test generation can reduce manual scripting time significantly, but integration costs, prompt engineering, and mandatory human review of generated tests add overhead. The cost per task-equivalent is approaching parity with human effort rather than offering clear savings.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply generate boilerplate test scripts, but human review, customization, and integration into existing systems still require significant paid engineering time, keeping costs roughly comparable for complete routines.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products like GitHub Copilot, ChatGPT, and specialized test generation tools (e.g., Diffblue, Testim) can draft test routines and scripts with demonstrated reliability in narrow domains. Material limitations remain in understanding nuanced requirements and validating coverage, so production use is common but still requires review.
Technical feasibility today	claude-sonnet-5	3/5	Coding assistants and AI test-generation tools (e.g., Copilot, testing frameworks with AI features) are deployed in production but typically handle unit-level or narrow test generation rather than full end-to-end testing strategy design.

Collaborate with Web developers to create and operate internal and external Web sites, or to manage projects, such as e-marketing campaigns.

CI 38–70 · exposure 45 · augmentation 88 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Technology and marketing-driven organizations have already adopted extensive automation for deployments, monitoring, and campaign management; pilot AI agents for operations are growing. This is a high-digitization sector with strong incentives and demonstrated deployment.
Sector adoption velocity	claude-sonnet-5	3/5	Web development and digital marketing are digitized fields with growing AI tool adoption (code assistants, content generators, analytics), though full project management remains human-led.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI systems routinely augment Web administrators through real-time monitoring dashboards, log analysis, automated alerting, campaign optimization suggestions, and A/B testing frameworks—substantially raising productivity while the human retains strategic oversight.
Augmentation potential	claude-sonnet-5	4/5	AI substantially aids coding, content drafting, campaign optimization, and analytics, meaningfully boosting productivity while humans still manage collaboration and decisions.
Task automatability	claude-haiku-4-5-20251001	4/5	Large portions of this task—from site deployment and configuration management to monitoring, A/B testing setup, and basic campaign analytics—can be automated or AI-assisted. However, strategic collaboration with developers and high-level project direction still require human judgment, so the remaining ~40–50% overhead prevents a full 5.
Task automatability	claude-sonnet-5	2/5	This is a collaborative, cross-functional task involving communication, project coordination, and stakeholder alignment that current AI cannot fully replace, though AI can accelerate coding, content generation, and campaign analytics sub-components.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Web operations and e-marketing are in digitized, information-intensive sectors with weak regulatory barriers. Organizational friction (need for cross-team sign-off) and reputational risk (broken campaigns) create mild friction, but no legal licensing or hard authorization requirement blocks automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational dependencies, stakeholder trust, and need for human judgment in cross-team collaboration create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	Infrastructure and marketing automation tools are now cost-effective compared to hiring dedicated staff; continuous deployment and analytics systems are mature and inexpensive. Full end-to-end replacement would require substantial setup, but marginal AI cost per incremental task is well below human wage.
Cost vs. human wage	claude-sonnet-5	2/5	Human collaboration, judgment, and coordination overhead remain necessary, so AI mainly reduces sub-task time rather than replacing the labor cost of the collaborative role itself.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed products exist for specific subtasks (CI/CD automation, infrastructure-as-code tools, marketing automation platforms), but end-to-end management of Web operations and cross-functional e-marketing campaigns involving real-time human collaboration remains patchy and error-prone at scale.
Technical feasibility today	claude-sonnet-5	2/5	AI coding assistants and marketing tools exist but no deployed product autonomously collaborates with human teams to manage full web projects or e-marketing campaigns end-to-end.

Determine sources of Web page or server problems, and take action to correct such problems.

CI 38–61 · exposure 38 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	High adoption is underway: IT/SRE teams widely deploy automated monitoring, alerting, and remediation (e.g., Kubernetes auto-recovery, CloudWatch auto-remediation, PagerDuty integrations). Information sector adoption is deep and measurable, with production displacement of routine diagnostics already common.
Sector adoption velocity	claude-sonnet-5	3/5	IT/ops sectors show moderate AI tool adoption (e.g., AIOps, copilots for sysadmins) but full autonomous remediation in production remains uncommon.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially augments human administrators through real-time dashboards, automated root-cause suggestions, and predictive alerting, enabling faster diagnosis and more efficient prioritization. The human remains in the loop for validation and complex fixes, with AI raising their throughput materially.
Augmentation potential	claude-sonnet-5	4/5	AI coding/debugging assistants and log-analysis tools meaningfully speed up root-cause identification and suggest fixes, significantly aiding administrators while they retain control.
Task automatability	claude-haiku-4-5-20251001	3/5	Partial automation is feasible: AI can diagnose common issues (log analysis, error pattern matching, basic connectivity checks) and suggest fixes, but complex root-cause analysis involving nuanced system interactions and novel failure modes still require human expertise. The task does not reach the 50% time-saving bar end-to-end for all problem types.
Task automatability	claude-sonnet-5	2/5	Diagnosing server/web issues requires investigating logs, live system state, and infrastructure-specific context that AI can assist with but not fully resolve end-to-end reliably today.ract
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few hard barriers exist; automation of diagnosis and routine fixes does not require licensing or legal sign-off. Customer preference for human validation and organizational caution around production system changes introduce some friction, but no regulatory or liability wall prevents substitution.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but production system changes carry real risk of downtime or data loss, creating organizational caution around fully automated fixes.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven monitoring and log analysis are cheap at scale (inference + integration costs are modest), while human web administrator wages are substantial. Automated systems can handle high-volume diagnostics for marginal cost once deployed, though complex investigations may still require expensive human time.
Cost vs. human wage	claude-sonnet-5	2/5	Complex troubleshooting still requires skilled human judgment and validation, so AI tools reduce but don't eliminate the labor cost, keeping costs roughly comparable to human effort for nontrivial issues.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed monitoring and diagnostic tools (e.g., APM platforms, automated alerting systems) exist and perform reliably for well-defined issues, but material limitations remain in handling ambiguous or multi-system failures. Production systems handle routine diagnostics but typically flag edge cases for human review.
Technical feasibility today	claude-sonnet-5	2/5	AI-assisted debugging tools and log analysis exist, but production systems rarely autonomously diagnose and fix arbitrary server/web problems without human oversight.

Test issues such as system integration, performance, and system security on a regular schedule or after any major program modifications.

CI 41–57 · exposure 50 · augmentation 88 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Large tech firms and financial services have deployed CI/CD automation and continuous testing extensively, but small-to-medium web hosting and administration shops still rely heavily on manual testing schedules. Adoption is uneven: mature DevOps shops use automated testing heavily, but human execution of formal test suites remains common in many organizations.
Sector adoption velocity	claude-sonnet-5	4/5	IT and software sectors have rapidly adopted automated testing, CI/CD, and security scanning tools as standard practice, reflecting fast adoption patterns typical of technical/professional services.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered testing tools (test generation, anomaly detection, automated report synthesis) substantially improve web administrator productivity by handling routine checks, freeing them to focus on investigating failures and designing complex test scenarios. Current tools transform efficiency on repetitive testing tasks while the human retains judgment over what to test and how to respond to findings.
Augmentation potential	claude-sonnet-5	5/5	AI-powered testing and security tools substantially boost administrator productivity by automating repetitive checks, generating test cases, and flagging anomalies, while humans retain final judgment and remediation responsibility.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can assist with generating test plans, executing automated security scans, and performance benchmarking; however, human judgment is still required for interpreting complex integration failures, risk prioritization, and deciding on remediation strategies. Partial automation of repetitive testing achieves moderate time savings but not full end-to-end replacement.
Task automatability	claude-sonnet-5	3/5	AI tools can generate and run test suites, scan for vulnerabilities, and flag performance regressions, but comprehensive system integration and security testing still requires human judgment for complex environments and edge cases.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory compliance (HIPAA, PCI-DSS, SOC 2) often mandates human sign-off on security testing and documentation by qualified personnel. Liability asymmetry means missed vulnerabilities or performance failures discovered post-deployment expose organizations to costly breaches, creating legal and organizational friction against full automation without human review.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement mandates a human perform this task, though organizational risk tolerance and the criticality of security failures create some friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Automated testing tools have significant upfront licensing and integration costs, plus require skilled humans to configure, tune, and interpret findings. While unit and regression testing automation is cheaper than manual testing, the full system testing and security validation tasks still require expensive engineer oversight, making the all-in cost competitive with or exceeding human testers.
Cost vs. human wage	claude-sonnet-5	3/5	Automated testing and scanning tools reduce labor costs substantially for routine checks, but licensing, tool maintenance, and required human review of results keep overall costs roughly comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed security scanning tools (SAST/DAST), performance monitoring platforms, and CI/CD testing automation exist in production, but they require human configuration, threshold-setting, and interpretation of results. False positives and the need for contextual judgment mean current products handle routine checks but struggle with novel integration scenarios.
Technical feasibility today	claude-sonnet-5	3/5	Deployed products (automated testing frameworks, CI/CD pipelines, security scanners like SAST/DAST tools) reliably handle parts of this task, but full end-to-end integration and security testing in production still involves significant human oversight and narrow tool scope.

Administer internet or intranet infrastructure, including Web, file, and mail servers.

CI 41–53 · exposure 50 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Mid-market and enterprise adoption of AI-assisted monitoring and patch automation is growing, but most organizations still rely on human-centric runbooks and change control boards. Adoption is concentrated in larger, digitized organizations; small-to-medium businesses lag significantly.
Sector adoption velocity	claude-sonnet-5	4/5	IT/infrastructure sectors have rapidly adopted automation, cloud-managed services, and AIOps tooling, representing one of the more digitized and fast-adopting domains.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at assisting web administrators by automating log analysis, alerting on anomalies, suggesting configurations, and accelerating routine tasks like deployment scripting. The human administrator remains essential for judgment calls, security decisions, and incident response, making this a high-augmentation scenario.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring, anomaly detection, automated scripting, and config generation substantially boost administrator productivity while humans retain oversight of critical infrastructure decisions.
Task automatability	claude-haiku-4-5-20251001	3/5	Routine server administration tasks such as monitoring, log analysis, basic troubleshooting, and configuration can be partially automated by AI-driven tools and agents, but complex infrastructure decisions, security incident response, and multi-system coordination still require human expertise and judgment. Roughly half the workload could be automated with significant setup and ongoing human oversight.
Task automatability	claude-sonnet-5	3/5	AI tools and scripts can automate significant portions of routine server administration (patching, config management, monitoring alerts), but incident diagnosis, architecture decisions, and security judgment still require human oversight, limiting full end-to-end automation.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory, liability, and organizational barriers exist: SOX/HIPAA/PCI-DSS compliance, audit trails, change management protocols, and vendor support contracts often legally require human accountability. Organizations are typically risk-averse about fully autonomous infrastructure changes due to potential downtime costs and data security implications.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement exists, but organizational risk aversion around security, uptime, and liability for infrastructure failures creates meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI monitoring and automation tools reduce some labor costs, but infrastructure administration requires low-latency, mission-critical response where human oversight and on-call availability remain expensive. The all-in cost of AI plus required human supervision approaches or exceeds the cost of specialized technicians for complex environments.
Cost vs. human wage	claude-sonnet-5	3/5	Automation tools reduce labor hours for routine maintenance tasks, but licensing, integration, and required human oversight for critical infrastructure keep costs roughly comparable to a lean human-plus-tooling team rather than an order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed products exist for infrastructure monitoring, automated patching, and log anomaly detection, but they operate with meaningful error rates and require human validation for critical actions. No current system reliably handles end-to-end autonomous administration of internet/intranet infrastructure without human sign-off on significant changes.
Technical feasibility today	claude-sonnet-5	3/5	Deployed products (IaC tools, AIOps monitoring, automated patch management, chatops) reliably handle subsets of infrastructure admin, but no product autonomously administers full web/file/mail server stacks without human sysadmins.

Perform user testing or usage analyses to determine Web sites' effectiveness or usability.

CI 32–60 · exposure 38 · augmentation 75 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Tech-forward organizations are adopting automated analytics and AI-assisted analysis tools, but full replacement of user testing remains limited; most adopt AI as a supplement to, not substitute for, human-led usability research.
Sector adoption velocity	claude-sonnet-5	3/5	Web administration and UX-related roles sit within tech/digital sectors with moderate AI adoption—analytics automation is common, but end-to-end AI-driven usability testing remains a pilot-stage practice in many organizations.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments the task by automating log parsing, generating heatmaps, identifying anomalies, and suggesting patterns from usage data, allowing human researchers to focus on interpretation, hypothesis formation, and deeper qualitative analysis rather than manual data collection.
Augmentation potential	claude-sonnet-5	4/5	AI tools significantly enhance a web administrator's ability to analyze usage patterns, generate reports, flag anomalies, and suggest UX improvements, meaningfully boosting productivity while humans still design and interpret tests.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can assist with automated testing, log analysis, and quantitative metrics gathering, but user testing fundamentally requires human subjects interacting with interfaces and qualitative interpretation of experience, which AI cannot replicate end-to-end at the required quality and depth.
Task automatability	claude-sonnet-5	3/5	AI can generate usage analyses, summarize analytics data, and even simulate user journeys or heuristic evaluations, but designing valid tests, recruiting/observing real users, and interpreting nuanced usability issues still require human judgment and setup.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Some organizational and methodological friction exists around accepting AI-driven insights in place of human user research, and domain expertise in UX is still valued, though no formal licensing or regulatory requirements mandate human-only execution.
Adoption barriers	claude-sonnet-5	1/5	There are no licensing or regulatory requirements mandating human-only usability testing; organizations are free to adopt automated tools without legal barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	While AI can reduce the cost of some analysis components (data aggregation, initial pattern detection), user testing still requires human researchers, participants, and judgment, making the total cost comparable to or higher than human-only approaches for equivalent insight.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted analytics tools reduce time spent on data processing and reporting, offering moderate cost savings, but comprehensive usability testing (recruiting users, moderated sessions) still requires human effort so overall cost parity is roughly comparable.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed tools exist for automated analytics, heatmap generation, and session recording, but the core task of determining effectiveness through user testing—observing user behavior, conducting interviews, and synthesizing insights—remains primarily human-dependent in production settings.
Technical feasibility today	claude-sonnet-5	3/5	Products exist (AI-powered analytics tools, heatmap/session analysis, automated usability testing platforms) that are used in production, but they typically supplement rather than replace human-led usability testing and often need configuration and interpretation.

Evaluate or recommend server hardware or software.

CI 36–55 · exposure 38 · augmentation 75 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	IT and technology sectors are moderately adopting AI-assisted tools for infrastructure recommendations, but full automation remains limited due to organizational risk aversion and the need to integrate with existing procurement workflows.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web administration functions are moderately fast adopters of AI tools for research and documentation, though infrastructure decision-making remains a slower-adopting niche within that sector.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments this task by rapidly analyzing specifications, benchmarking options, generating cost comparisons, and summarizing tradeoffs, which accelerates Web Administrator decision-making while they retain critical judgment on final recommendations.
Augmentation potential	claude-sonnet-5	4/5	AI is highly useful for summarizing hardware/software specs, drafting comparison matrices, and surfacing tradeoffs, meaningfully speeding up the evaluation process even though humans finalize decisions.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can assist in analyzing server specifications, comparing vendors, and generating recommendations based on documented requirements, but the task typically involves complex tradeoffs, legacy constraints, and organizational context that require human judgment to finalize.
Task automatability	claude-sonnet-5	2/5	AI can gather specs, summarize benchmarks, and draft comparisons, but final evaluation involves judgment about organizational context, cost negotiation, and integration risk that current AI cannot fully own end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Organizations often require human sign-off on hardware and software decisions for liability and compliance reasons, and procurement policies frequently mandate vendor engagement; however, these are procedural rather than legal barriers.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement exists, but organizational risk aversion around infrastructure decisions and accountability for costly hardware/software choices creates moderate friction against pure AI-driven recommendations.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI-powered analysis tools cost significantly less than hiring consultants, but Web Administrators still command moderate salaries; the cost is roughly comparable when accounting for oversight and integration overhead.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted research can cut time spent gathering comparative data significantly, but human validation, vendor negotiation, and infrastructure-specific testing still require substantial paid expert time, keeping costs roughly comparable.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products like configuration analysis tools and AI-assisted recommendation systems exist and are used in production, but they often require significant human validation and don't achieve full autonomous end-to-end evaluation without oversight.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and copilots are used informally for research and comparison drafting, but no deployed product independently evaluates and recommends server infrastructure reliably in production workflows.

Develop and implement marketing plans for home pages, including print advertising or advertisement rotation.

CI 30–57 · exposure 33 · augmentation 75 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While marketing teams use AI writing and design tools, actual adoption of AI-driven autonomous marketing plan development and implementation remains limited; most organizations still rely on human strategists to own and drive these decisions in production.
Sector adoption velocity	claude-sonnet-5	4/5	Marketing and digital advertising are among the fastest-adopting sectors for AI tools, with widespread production use of AI for content generation, ad optimization, and campaign planning.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments marketing plan development by rapidly generating ad copy variants, suggesting design layouts, automating A/B testing analysis, and recommending advertisement rotation strategies. Human marketers using these tools can operate substantially faster while maintaining strategic control.
Augmentation potential	claude-sonnet-5	4/5	AI significantly boosts productivity for brainstorming campaign ideas, drafting ad copy, analyzing performance data, and generating rotation schedules, while humans retain strategic control.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate advertising copy and suggest layout options, developing and implementing complete marketing plans requires strategic positioning, brand alignment, and campaign management decisions that currently demand human oversight. AI can assist with individual components (ad copy, design suggestions) but cannot end-to-end execute the full planning and implementation cycle.
Task automatability	claude-sonnet-5	3/5	AI can draft marketing plan content, ad copy, and rotation schedules, but strategic decisions about brand positioning, budget allocation, and market fit still require human judgment and oversight to reach equal quality output.'
Adoption barriers	claude-haiku-4-5-20251001	3/5	Marketing plans require business judgment and brand strategy knowledge; organizations typically prefer human marketers for accountability and creative direction. However, there are no hard legal or regulatory barriers preventing automation of the technical execution components.
Adoption barriers	claude-sonnet-5	1/5	Marketing planning carries no licensing or legal sign-off requirements, and there is no regulatory or human-contact barrier preventing AI-assisted or AI-driven marketing plan development.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI tools (copywriting, design assistance) cost money and still require significant human marketing oversight, strategy, and execution oversight, making the all-in cost comparable to or higher than a junior marketer performing the task.
Cost vs. human wage	claude-sonnet-5	3/5	AI tools reduce time spent on drafting and ideation substantially, but human strategists, designers, and analysts are still needed for implementation and oversight, keeping costs roughly comparable when factoring in integration and review.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No production system reliably performs end-to-end marketing plan development and implementation autonomously. Tools exist for ad copy generation and A/B testing, but integrating these into coherent strategic plans with print and digital coordination remains manual and requires human marketing expertise.
Technical feasibility today	claude-sonnet-5	2/5	Marketing copilots and content generation tools exist and are used for drafting, but no deployed product reliably develops and implements a full home page marketing plan end-to-end without significant human direction.

Test new software packages for use in Web operations or other applications.

CI 32–51 · exposure 38 · augmentation 75 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	DevOps and web operations teams are actively adopting CI/CD and automated testing, but AI-specific test generation and autonomous test management remain in pilot phase; mainstream adoption is emerging but not yet deep.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web development sectors show above-average AI tool adoption for coding and testing assistance, though full autonomous software vetting remains uncommon in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists by generating test cases, suggesting edge cases, automating repetitive regression testing, and providing detailed failure analysis—allowing human testers to focus on exploratory and security testing while staying in decision-making oversight.
Augmentation potential	claude-sonnet-5	4/5	AI coding assistants and testing tools significantly speed up tasks like writing test scripts, identifying bugs, and summarizing compatibility issues, meaningfully boosting administrator productivity.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can assist with automated testing (unit tests, regression suites, basic functional testing) and generate test cases, achieving partial time savings; however, exploratory testing, security validation, and integration testing with existing systems require domain knowledge and human judgment.
Task automatability	claude-sonnet-5	2/5	Testing new software for web operations requires hands-on evaluation of compatibility, security, performance, and integration in real environments, which AI can assist but not fully replace given the judgment and contextual decision-making needed.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Web operations testing often requires sign-off on deployment safety and production validation, creating organizational friction; liability concerns around failed deployments create incentive for human verification, though testing itself is not legally restricted to licensed professionals.
Adoption barriers	claude-sonnet-5	2/5	No strict licensing requirement exists, but organizational risk tolerance and the need for accountable sign-off on production software changes create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-driven test automation tools reduce repetitive manual testing costs, but setup, integration, and oversight overhead—plus the skilled engineer time needed to configure and validate—keep total costs comparable or higher than focused human testing.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can reduce some testing labor via automated scripts, human oversight, infrastructure setup, and validation still dominate costs, keeping AI only modestly cheaper at best.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products exist (automated testing frameworks, AI-assisted test generation tools like GitHub Copilot) that perform parts of this task reliably, but they typically cover narrow testing domains and require significant human-driven test planning and result interpretation.
Technical feasibility today	claude-sonnet-5	2/5	AI tools can help generate test cases, scan for vulnerabilities, or summarize documentation, but no deployed product autonomously performs full software evaluation and adoption decisions for web operations.

Monitor Web developments through continuing education, reading, or participation in professional conferences, workshops, or groups.

CI 29–46 · exposure 17 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Many tech organizations use AI-powered news aggregators and summaries (e.g., curated tech feeds, ChatGPT summaries of conference materials), but adoption remains supplementary rather than replacing human participation in professional development and networking.
Sector adoption velocity	claude-sonnet-5	3/5	Tech workers broadly use AI tools like summarizers and chatbots for staying current, a middling-to-fast adoption pattern within IT/software sectors.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at surfacing relevant articles, summarizing conference talks, and organizing technical news feeds, significantly amplifying a web admin's ability to stay current without attending every event or reading every source. This assistive role meaningfully boosts productivity while the human retains decision-making.
Augmentation potential	claude-sonnet-5	4/5	AI can effectively summarize articles, aggregate trending topics, and answer questions about new web technologies, meaningfully speeding up a professional's ability to stay current.
Task automatability	claude-haiku-4-5-20251001	1/5	This task fundamentally requires human judgment to evaluate relevance, synthesize information, and decide which developments matter for specific organizational contexts. Current AI cannot autonomously identify which conferences to attend or which professional discussions merit attention without explicit human direction.
Task automatability	claude-sonnet-5	2/5	AI can summarize articles and curate feeds, but genuinely 'monitoring developments' through professional engagement, judgment about relevance, and networking requires human initiative and cannot be fully offloaded to AI.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Professional development and continuing education are often part of employment expectations or compliance requirements where human judgment and discretion are valued. Some organizations have explicit policies requiring staff to engage directly with professional communities, though barriers are not strict legal requirements.
Adoption barriers	claude-sonnet-5	1/5	No licensing, regulatory, or human-contact requirement blocks using AI tools to assist with staying current on web technology trends.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools (newsletters, summarization, aggregation services) are inexpensive, but the human reading, filtering, and interpreting time remains substantial. The cost of AI assistance is low but doesn't eliminate the need for human engagement, making the all-in ratio unfavorable for full automation.
Cost vs. human wage	claude-sonnet-5	3/5	AI summarization tools are cheap relative to time spent reading, but this task is not a discrete billable output so cost comparison is muddled; savings are moderate rather than transformative.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can summarize published articles or conference agendas at scale, no deployed product reliably performs the full task of selecting, filtering, and contextualizing web developments for a specific admin's needs. Systems lack the domain judgment to distinguish signal from noise in rapidly evolving technical landscapes.
Technical feasibility today	claude-sonnet-5	2/5	Tools like AI-curated newsletters and summarization assistants exist and are used informally, but no deployed product autonomously performs professional development monitoring for a worker in production.

Evaluate testing routines or procedures for adequacy, sufficiency, and effectiveness.

CI 32–41 · exposure 25 · augmentation 75 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Tech companies are actively adopting AI-assisted testing tools and continuous integration enhancements, but production replacement of human evaluation judgment remains limited; mostly pilots and augmentation in practice.
Sector adoption velocity	claude-sonnet-5	3/5	Software/IT sectors have moderate-to-fast AI adoption for code analysis and testing tools, though full evaluation workflows still rely heavily on human engineers.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can substantially assist web administrators by auto-generating test suggestions, identifying coverage gaps, and flagging potential failure scenarios, meaningfully raising their evaluation productivity while they retain final judgment.
Augmentation potential	claude-sonnet-5	4/5	AI tools can significantly speed up identifying test coverage gaps, generating test cases, and flagging weaknesses, greatly aiding a human evaluator's process.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can generate test scripts and identify missing coverage patterns in existing test suites, but evaluating adequacy and effectiveness requires domain knowledge, understanding business context, and judgment about risk trade-offs that current systems struggle with reliably.
Task automatability	claude-sonnet-5	2/5	Evaluating testing routines requires contextual judgment about coverage, edge cases, and business risk that AI can assist with but not fully replace end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While testing evaluation is not legally mandated in most jurisdictions, organizational practices, liability concerns for critical systems, and the need for expert sign-off create moderate friction against full automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational trust and accountability for system reliability create moderate friction against fully automating this judgment call.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-powered testing analysis tools are relatively affordable, but web administrators still require substantial oversight and validation time, making the all-in cost comparable to or higher than direct human evaluation.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted static analysis and coverage tools are cheap to run, but the human review needed to validate sufficiency and effectiveness keeps overall cost roughly comparable to human-only review.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Tools exist to analyze test coverage metrics and suggest improvements, but no deployed products reliably and independently evaluate whether testing procedures are truly sufficient for production systems without significant human expert review.
Technical feasibility today	claude-sonnet-5	2/5	AI code-review and test-analysis tools exist and can flag gaps in test coverage, but no deployed product reliably performs holistic evaluation of testing adequacy without human oversight.

Collaborate with development teams to discuss, analyze, or resolve usability issues.

CI 32–38 · exposure 25 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Technology firms are experimenting with AI-assisted usability analysis and bug detection, but actual deployment in collaborative workflows remains in the pilot stage rather than widespread production use.
Sector adoption velocity	claude-sonnet-5	3/5	Software/web development teams are fast adopters of AI coding and analysis tools, though the specific collaborative discussion component is less directly automated, giving middling overall adoption for this specific task.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at augmenting this task by automatically flagging usability issues from logs, summarizing user feedback, and generating analysis drafts that developers then discuss and refine—substantially boosting human analyst productivity while keeping humans central to decision-making.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by pre-analyzing usability data, drafting summaries, generating heatmaps/reports, and suggesting fixes, boosting productivity while humans still lead the collaboration.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in analyzing usability data and generating initial insights, the collaborative discussion aspect and human judgment required to resolve complex usability issues require significant human participation. AI cannot fully replace the back-and-forth negotiation and contextual understanding typical of cross-team collaboration.
Task automatability	claude-sonnet-5	2/5	This is a collaborative, real-time discussion task requiring judgment, stakeholder alignment, and contextual understanding of business needs; AI can support analysis but cannot conduct the collaboration itself end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While there are no strict licensing barriers, organizational friction and preference for human domain expertise in technical collaboration present moderate adoption friction. Teams often distrust automated suggestions without human vouching.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational and interpersonal dynamics of cross-team collaboration create moderate friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Deploying AI systems with sufficient oversight and integration to meaningfully participate in team discussions remains expensive relative to a junior web administrator's hourly cost, especially when human review and correction are necessary.
Cost vs. human wage	claude-sonnet-5	2/5	Human meetings and negotiation still require paid staff time; AI can reduce prep/analysis time but doesn't replace the collaborative labor itself, so cost savings are modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools can analyze usability logs and suggest issues, but no deployed product reliably conducts end-to-end collaborative problem-solving with development teams at production scale. Current systems lack the nuanced communication and real-time adaptation needed for effective team collaboration.
Technical feasibility today	claude-sonnet-5	2/5	AI tools (chatbots, code analysis, usability heuristics checkers) can flag issues, but no deployed product autonomously runs cross-team usability discussions or resolves disputes in production.

Recommend Web site improvements, and develop budgets to support recommendations.

CI 32–38 · exposure 25 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Website monitoring and performance tools are widely adopted in digitized organizations, but adoption of AI-driven recommendation engines for strategic improvements and budgeting remains in the pilot or limited-production phase; this is not yet a standard workflow.
Sector adoption velocity	claude-sonnet-5	3/5	Web administration sits within IT/digital services, a moderately fast-adopting sector, though budget planning tasks are less automated than technical execution tasks.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI tools can substantially assist web administrators by rapidly analyzing site performance data, surfacing optimization opportunities, and generating initial cost estimates, allowing the human to focus on strategic judgment and stakeholder alignment rather than manual data gathering.
Augmentation potential	claude-sonnet-5	4/5	AI tools can analyze site metrics, competitor benchmarks, and cost data to strongly support drafting improvement plans and budget justifications, significantly speeding up the human's work.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze website data and suggest surface-level improvements (e.g., performance metrics, accessibility issues), recommending substantive improvements requires understanding business context, user needs, and strategic priorities that demand human judgment. Budget development requires cost estimation and organizational alignment that current systems cannot reliably do end-to-end.
Task automatability	claude-sonnet-5	2/5	AI can generate suggestions and draft budget estimates, but synthesizing site analytics, business priorities, and stakeholder needs into actionable recommendations requires human judgment that current AI cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	No formal licensing or regulatory barrier exists, but organizational friction is moderate: recommendations must align with business strategy, stakeholder buy-in is required, and liability for poor budget decisions or failed improvements creates pressure for human accountability and review.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but budget approval and strategic recommendations typically require organizational accountability and sign-off from someone with institutional authority.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for website analysis exist at low per-use cost, but the human judgment, stakeholder interviews, and strategic planning that dominate this task remain expensive. Integration and oversight costs to validate AI-generated recommendations would likely approach or exceed the value of lightweight automation.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply generate draft suggestions, but the human oversight needed to validate recommendations and construct credible budgets against organizational constraints keeps overall cost comparable to human effort.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Products exist for website analytics and optimization suggestions (Lighthouse, PageSpeed Insights), but they address narrow aspects of the task. No deployed product reliably generates comprehensive, business-contextualized improvement recommendations with accompanying justification and budgets that a web administrator could execute directly.
Technical feasibility today	claude-sonnet-5	2/5	Analytics tools and AI assistants can surface UX/performance issues and suggest improvements, but no deployed product reliably produces vetted improvement roadmaps with accurate budget justifications in production.

Develop or implement procedures for ongoing Web site revision.

CI 32–38 · exposure 25 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Website operations and IT infrastructure teams are moderately digitized and have access to AI tools, but actual adoption of AI for designing revision procedures in production is still in the pilot phase, not yet widespread replacement behavior.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web operations teams increasingly use AI tools for documentation and workflow drafting, but full procedural automation adoption is still emerging and inconsistent.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by drafting procedure templates, suggesting revision workflows, identifying potential gaps in existing procedures, and generating documentation—raising a web administrator's productivity in this task while the human retains final authority and decision-making.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully help draft, structure, and refine revision procedures, checklists, and documentation, significantly speeding up the administrator's planning work.
Task automatability	claude-haiku-4-5-20251001	2/5	Website revision procedures involve significant domain knowledge, business logic decisions, and context-dependent judgment about what changes are needed and when. While AI can draft procedures or suggest templates, developing comprehensive, organization-specific revision workflows that account for stakeholder coordination, testing protocols, and rollback strategies requires human oversight and cannot meet the 50% time-saving threshold end-to-end.
Task automatability	claude-sonnet-5	2/5	Developing governance procedures for ongoing site revision requires organizational judgment, stakeholder alignment, and policy design that current AI can assist with but not fully execute end-to-end.ed AI can draft procedure documents but cannot independently establish and implement organizational workflows.
Adoption barriers	claude-haiku-4-5-20251001	3/5	There are moderate barriers to full automation: procedures typically require sign-off from IT leadership or security teams, liability for failed rollouts rests with humans, and organizational friction around delegating strategic procedural design to automated systems is substantial. However, no strict licensing requirement prevents an AI system from assisting or augmenting this work.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational change management, internal politics, and accountability for site changes create moderate friction against pure automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Web administrators' loaded costs ($50–80K/year equivalent) are moderate, and while AI tools can reduce time on drafting procedures, the supervision and refinement required for production-grade procedures means the all-in AI cost per completed task remains comparable to or exceeds human cost.
Cost vs. human wage	claude-sonnet-5	2/5	Human process owners still need to gather requirements, negotiate stakeholder buy-in, and enforce procedures, so AI only reduces drafting time rather than the bulk of labor cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature products reliably perform the full task of developing and implementing custom website revision procedures in production. AI can assist with documentation generation and suggest best practices, but the procedural design itself—accounting for organizational constraints, risk tolerance, and specific tech stacks—remains largely manual.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously creates and implements web revision governance procedures within an organization; this remains a human-led process management task with AI as a drafting aid.

Identify, standardize, and communicate levels of access and security.

CI 28–37 · exposure 30 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Mature adoption of identity and access management platforms is common in larger organizations, but true automation of access policy definition and communication remains at the pilot stage; most companies still rely primarily on manual review and human decision-making.
Sector adoption velocity	claude-sonnet-5	3/5	IT and security functions are moderately fast adopters of AI tooling, with pilots for access management automation common, but full production reliance is still limited.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI tools effectively augment this task by scanning infrastructure, generating access reports, flagging anomalies, and suggesting standardization templates, while administrators make final policy decisions and communicate changes. This assistive layer meaningfully raises administrator productivity without removing human oversight.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully assist by drafting access policies, flagging inconsistencies, and generating documentation, substantially speeding up the human-led standardization process.
Task automatability	claude-haiku-4-5-20251001	2/5	Identifying and standardizing access levels involves policy definition, organizational decision-making, and contextual judgment that AI cannot fully automate. AI can assist with scanning systems to document current access, but the core task of determining appropriate security levels requires human expertise and organizational authority.
Task automatability	claude-sonnet-5	2/5	This task involves policy judgment, organizational context, and communication with stakeholders that current AI cannot fully replace, though it can draft access-control matrices or policy templates.dah
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security and access control are heavily regulated (compliance frameworks, governance requirements, audit trails). Most organizations require human security architects or administrators to define, approve, and sign off on access policies for liability and audit purposes.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement, but liability for security misconfiguration and organizational trust requirements create meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	While scanning and documentation can be partially automated at low cost, the human overhead for policy review, approval, and customization remains high. The total cost of AI-assisted tooling plus required human oversight is often comparable to or exceeds direct human effort.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply generate draft policy documents, but the human review, negotiation, and organizational rollout components remain costly, keeping overall cost comparable to human-led effort.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products exist for access governance and security auditing (e.g., Okta, CyberArk, identity management platforms), but they require significant human configuration, policy input, and validation. Fully autonomous policy definition and communication without human oversight remains unreliable.
Technical feasibility today	claude-sonnet-5	2/5	Some products (IAM tools, AI-assisted security platforms) suggest access policies, but no deployed product independently identifies, standardizes, and communicates security access levels across an organization without human oversight.

Implement Web site security measures, such as firewalls or message encryption.

CI 25–37 · exposure 30 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Despite high digital infrastructure adoption, actual automation of security implementation remains limited in production environments due to risk aversion and the human-accountable nature of security decisions. Most organizations maintain human-led security architecture even where AI assists.
Sector adoption velocity	claude-sonnet-5	3/5	IT/security is a moderately fast-adopting sector for AI copilots in coding and threat detection, but full autonomous implementation of security measures remains rare in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI provides substantial augmentation through automated vulnerability scanning, configuration recommendations, threat detection, and policy suggestion—materially raising human administrator productivity. Humans remain responsible for final decisions, but AI transforms their capability to analyze and recommend.
Augmentation potential	claude-sonnet-5	4/5	AI tools meaningfully assist by suggesting configurations, generating boilerplate encryption/firewall code, and flagging vulnerabilities, while humans retain decision-making and validation.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in recommending security configurations and identifying vulnerabilities, implementing actual security measures requires human judgment about organizational architecture, compliance requirements, and risk tolerance. Current AI cannot reliably execute end-to-end implementation with 50% time savings at equal quality given the criticality and context-specific nature of security decisions.
Task automatability	claude-sonnet-5	2/5	Configuring firewalls and encryption involves architectural judgment, risk assessment, and context-specific decisions that current AI can assist with but not reliably execute end-to-end without expert oversight.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security implementation faces strong barriers: regulatory compliance requirements (PCI-DSS, HIPAA, SOC 2), liability exposure if AI-configured systems fail, organizational risk management protocols, and explicit authorization requirements in many sectors. Humans are typically legally accountable for security posture.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement exists, but liability for breaches and compliance obligations (PCI-DSS, GDPR) create strong incentives for human accountability and sign-off.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI security tools are available but typically require significant human oversight and customization; the all-in cost (including human validation, integration, and liability management) remains comparable to or higher than direct human implementation for most organizations.
Cost vs. human wage	claude-sonnet-5	2/5	Security misconfigurations carry high remediation and breach costs, so human expert review remains necessary, keeping AI-only cost savings limited despite cheap inference.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Security tools with AI-assisted features (vulnerability scanning, policy recommendation) exist and are deployed, but most implementation still requires human oversight, configuration, and validation. No mature product fully automates security infrastructure implementation end-to-end.
Technical feasibility today	claude-sonnet-5	2/5	AI coding assistants can generate config snippets or suggest security settings, but no deployed product autonomously implements and validates full site security architecture in production.

Test backup or recovery plans regularly and resolve any problems.

CI 25–37 · exposure 30 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While backup automation is widespread, systematic AI-driven testing and remediation of recovery plans remains rare in production; most organizations still rely on scheduled manual testing or basic alerting rather than autonomous problem resolution.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations broadly adopt automation and monitoring tools at a moderate pace, with backup automation common but full autonomous recovery testing and issue resolution still emerging as pilots rather than standard practice.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist by automating test execution, parsing logs for anomalies, suggesting remediation steps, and generating compliance reports—allowing administrators to focus on validation and critical decisions rather than repetitive manual testing.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring, anomaly detection, and scripted testing tools significantly assist administrators in scheduling tests, flagging failures, and suggesting fixes, meaningfully boosting productivity while humans retain final control.
Task automatability	claude-haiku-4-5-20251001	2/5	Testing backup/recovery plans requires execution of structured steps that AI can partially automate (running test scripts, logging results), but resolving problems discovered during testing demands human judgment, configuration decisions, and understanding of business criticality that current AI systems cannot reliably handle without extensive oversight.
Task automatability	claude-sonnet-5	2/5	Executing backup/recovery test scripts can be automated, but diagnosing and resolving unexpected failures requires human judgment about system architecture and business impact, limiting end-to-end automation today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Recovery plan testing often has implicit organizational and regulatory requirements (audit trails, sign-off on recovery RTO/RPO validation) and carries high liability costs if failures go undetected; many organizations require human certification of backup integrity.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement, but organizational risk aversion around data loss and system outages creates strong preference for human verification and sign-off before trusting automated recovery validation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Automating test execution reduces labor but the critical problem-resolution phase still requires skilled web administrators; the all-in cost of AI orchestration plus human oversight for failures likely exceeds simple human performance of the task.
Cost vs. human wage	claude-sonnet-5	2/5	While automated scripts reduce some routine checking cost, troubleshooting failures and validating recovery integrity still requires skilled human oversight, keeping all-in AI cost close to or above human cost for full task completion.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Some backup testing automation exists (scheduling, log parsing, alerting on failures), but deployed products still require significant human intervention to diagnose root causes and validate recovery procedures in production contexts.
Technical feasibility today	claude-sonnet-5	2/5	Some monitoring and backup-verification tools exist (e.g., automated backup validation), but comprehensive disaster-recovery testing and problem resolution in production environments still rely heavily on human sysadmins.

Identify or address interoperability requirements.

CI 25–38 · exposure 25 · augmentation 50 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Web administration and IT infrastructure sectors show slow adoption of AI agents for architectural decisions. Most organizations still rely on human expertise for interoperability planning; adoption remains at the pilot stage rather than production deployment for autonomous decision-making.
Sector adoption velocity	claude-sonnet-5	3/5	IT and web administration functions are in a moderately fast-adopting sector, with AI coding/config assistants seeing growing pilot and production use, though this specific requirements-analysis task lags behind coding automation.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by analyzing system documentation, suggesting potential integration points, and flagging compatibility conflicts. However, the human administrator must ultimately validate requirements against business context, making this an assistance tool rather than a productivity transformer.
Augmentation potential	claude-sonnet-5	3/5	AI can help research standards, compare API specs, draft compatibility documentation, and suggest potential integration issues, meaningfully aiding but not replacing the administrator's analysis.
Task automatability	claude-haiku-4-5-20251001	2/5	Identifying interoperability requirements requires understanding complex system architectures, business needs, and technical constraints. While AI can assist in analyzing logs and documentation, the task involves significant judgment about organizational goals and cross-system dependencies that typically demands human expertise and stakeholder input.
Task automatability	claude-sonnet-5	2/5	Identifying interoperability requirements requires understanding diverse systems, stakeholder needs, and architectural context that AI cannot fully assess autonomously today; it can assist but not replace the end-to-end judgment involved.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Significant organizational and liability barriers exist: interoperability decisions typically require sign-off from IT architects or stakeholders, affect multiple systems and departments, and carry risk of system failures if incorrect. Regulatory compliance (data access, security) often mandates human accountability for these requirements.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational friction exists since interoperability decisions affect multiple systems/teams and typically require sign-off from technical leads.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The setup, integration, and continuous oversight costs of AI systems for this task, plus inevitable human validation of architectural decisions, exceed what a skilled web administrator would cost for the same work. The high cost of errors in interoperability decisions adds substantial verification overhead.
Cost vs. human wage	claude-sonnet-5	2/5	Because human architectural judgment and cross-team communication are still required, AI tools reduce some research time but don't replace the bulk of the labor cost involved.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably perform independent interoperability requirement identification at enterprise scale. AI tools can help document existing systems or flag potential conflicts, but they lack the contextual understanding and authority to authoritatively identify and address requirements without substantial human oversight and validation.
Technical feasibility today	claude-sonnet-5	2/5	Some AI coding/documentation assistants can flag compatibility issues in code or configs, but no deployed product reliably identifies or resolves cross-system interoperability requirements as a standalone capability.

Related occupations — Computer & Mathematical

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.