Network and Computer Systems Administrators

15-1244.00

Median wage $99,130/yr314,340 employed (US)Rank #88 of 923 scored · top 10% by substitution

Install, configure, and maintain an organization's local area network (LAN), wide area network (WAN), data communications network, operating systems, and physical and virtual servers. Perform system monitoring and verify the integrity and availability of hardware, network, and server resources and systems. Review system and application logs and verify completion of scheduled jobs, including system backups. Analyze network and server resource consumption and control user access. Install and upgrade software and maintain software licenses. May assist in network modeling, analysis, planning, and coordination between network and data communications hardware and software.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution49

Exposure44

Augmentation70

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

20 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

15%

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%43

panel mean rating 2.7/5 → substitution pressure 43/100

Technical feasibility todayw 20%45

panel mean rating 2.8/5 → substitution pressure 45/100

Cost vs. human wagew 15%46

panel mean rating 2.9/5 → substitution pressure 46/100

Adoption barriersw 20%inverted — strong barriers lower the score59

panel mean rating 2.6/5 (barrier strength) → substitution pressure 59/100

Sector adoption velocityw 10%54

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

Task breakdown (20 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.

Perform routine network startup and shutdown procedures, and maintain control records.

CI 75–95 · exposure 87 · augmentation 75 · importance 3.4/5 · 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	Enterprise IT, cloud infrastructure, and DevOps teams have already embedded automation into standard operations; startup/shutdown automation is now standard practice in tech-forward organizations and increasingly expected in regulated sectors.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and network management have seen fast adoption of automation and orchestration tools, driven by DevOps and NOC efficiency initiatives.
Augmentation potential	claude-haiku-4-5-20251001	4/5	Automation directly augments sysadmin productivity by eliminating manual, repetitive execution, freeing them to focus on exception handling, monitoring, and strategic infrastructure changes while maintaining visibility into control records.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring and automation significantly reduce manual effort and error in routine procedures while administrators retain oversight and handle exceptions.
Task automatability	claude-haiku-4-5-20251001	5/5	Routine startup and shutdown procedures are highly structured, repeatable sequences with clear prerequisites and state checks; modern infrastructure-as-code and automation tools (Ansible, Terraform, cloud APIs) can execute these end-to-end with >50% time savings and equivalent or better reliability than manual execution.
Task automatability	claude-sonnet-5	4/5	Routine startup/shutdown procedures and record-keeping are highly scriptable and can be handled by automation tools, scheduled scripts, and monitoring systems with minimal human intervention.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While organizations may require human oversight for change control and regulatory compliance documentation, the actual automation itself faces minimal legal or licensing barriers; IT governance policies (not technical requirements) are the primary friction.
Adoption barriers	claude-sonnet-5	2/5	No licensing or legal requirement mandates a human perform this; main friction is internal change-management policies and ensuring reliability before full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated startup/shutdown via cloud APIs or self-hosted orchestration costs pennies per execution versus the loaded hourly wage of a sysadmin ($50–100+/hour), representing at least a 100× cost advantage.
Cost vs. human wage	claude-sonnet-5	4/5	Once configured, automated scripts and monitoring tools run at a fraction of the ongoing labor cost compared to manual execution and logging by an administrator.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Production-grade automation and orchestration platforms (AWS Systems Manager, Kubernetes controllers, enterprise configuration management) reliably handle network startup/shutdown at scale across thousands of organizations today, with mature logging and state management.
Technical feasibility today	claude-sonnet-5	4/5	Network management platforms (e.g., Ansible, SolarWinds, orchestration tools) already perform automated startup/shutdown sequencing and log control records reliably in production environments.

Perform data backups and disaster recovery operations.

CI 80–86 · exposure 84 · 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	5/5	Backup and disaster recovery automation has been adopted at scale in enterprise IT for over a decade; cloud and managed services have accelerated this further, with wide deployment across finance, healthcare, and corporate sectors.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and infrastructure management are among the most digitized functions with widespread adoption of automated backup and DR tooling across most mid-to-large organizations already.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI assists administrators by automating routine operations, alerting on failures, suggesting recovery strategies, and validating backup integrity, substantially raising productivity while administrators handle exception cases and policy decisions.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring, anomaly detection, and automated runbooks significantly boost administrator productivity in managing backup schedules and executing recovery procedures, while humans retain oversight for complex incident response.
Task automatability	claude-haiku-4-5-20251001	4/5	Modern backup and disaster recovery systems are highly automated with scheduling, incremental backups, and failover orchestration built-in. AI can monitor, trigger, and coordinate most operations, though complex recovery decisions and validation typically require human judgment, limiting it to ~70–80% of routine time savings.
Task automatability	claude-sonnet-5	4/5	Modern backup software already automates scheduling, execution, verification, and failover with minimal human intervention, and cloud-native DR orchestration tools can automate most of the routine workflow, though complex disaster scenarios still need human judgment for recovery planning and validation.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While backups are critical and organizations may prefer human oversight for verification, there are minimal legal or licensing barriers preventing automation; most barriers are operational (testing, validation workflows) rather than regulatory.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement mandates human execution, but organizational risk tolerance and compliance/audit requirements (e.g., verifying recovery integrity) create moderate friction before fully unattended automation is trusted.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated backup infrastructure and AI-driven orchestration cost a small fraction of hiring and paying a full-time administrator for continuous backup oversight, representing an order-of-magnitude savings.
Cost vs. human wage	claude-sonnet-5	4/5	Automated backup/DR platforms handle vast volumes of data operations at a fraction of the cost of manual administration, though licensing, storage, and oversight costs keep it from being a full order-of-magnitude cheaper in all cases.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature enterprise backup solutions (Veeam, NetBackup, Commvault, AWS/Azure native tools) perform backups and recovery operations reliably in production at scale across millions of organizations daily.
Technical feasibility today	claude-sonnet-5	5/5	Mature enterprise products (Veeam, AWS Backup, Azure Site Recovery, Rubrik, Commvault) reliably perform automated backups and orchestrated disaster recovery in production environments at scale today.

Maintain logs related to network functions, as well as maintenance and repair records.

CI 75–91 · exposure 80 · 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	5/5	Automated log management is deeply embedded in IT infrastructure across enterprise, finance, cloud, and tech sectors; this is among the most widely deployed automation patterns in network administration.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and network administration sectors have rapidly adopted automated logging, monitoring, and record-keeping tools as standard practice, reflecting fast, deep adoption typical of tech-forward professional environments.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-assisted log analysis and anomaly detection can alert administrators to issues, suggest categorization, and accelerate root-cause investigation, significantly boosting human productivity while keeping the administrator in decision-making roles.
Augmentation potential	claude-sonnet-5	4/5	AI-enhanced log analysis and automated documentation tools significantly boost administrators' ability to track, summarize, and act on maintenance data, though human review remains valuable for context and troubleshooting decisions.
Task automatability	claude-haiku-4-5-20251001	4/5	Log maintenance and record-keeping are highly structured, repetitive tasks amenable to automation. Current AI systems can parse, organize, categorize, and archive logs with minimal human oversight, and automated monitoring tools already perform these functions in production, achieving substantial time savings.
Task automatability	claude-sonnet-5	4/5	Log maintenance and record-keeping is largely structured data entry and aggregation, which current AI and automation tools (log management systems, scripts, AI-assisted documentation) can handle with high time savings, though some contextual annotation may still need human input.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While some regulatory frameworks (SOC 2, HIPAA) require audit trails and retention policies, the task itself is not legally restricted to human operators. The main friction is organizational preference for human oversight and vendor lock-in, not hard legal barriers.
Adoption barriers	claude-sonnet-5	1/5	There is no licensing, regulatory, or liability barrier to automating log and maintenance record-keeping; it's a routine administrative function.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated log management and archival systems cost orders of magnitude less than hiring staff to manually review, categorize, and file logs; once deployed, marginal cost per log processed is negligible.
Cost vs. human wage	claude-sonnet-5	4/5	Automated logging and record systems are inexpensive relative to manual documentation labor, though some integration and oversight costs remain, keeping it just short of order-of-magnitude savings in all contexts.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature products (SIEM platforms, log aggregation tools, automated ticket systems) reliably perform log collection, parsing, and record maintenance at scale in production environments across thousands of organizations.
Technical feasibility today	claude-sonnet-5	4/5	Mature log management and IT service management products (e.g., Splunk, ServiceNow, SolarWinds) already automate log capture, aggregation, and record-keeping in production environments at scale.

Analyze equipment performance records to determine the need for repair or replacement.

CI 55–79 · exposure 62 · 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	Enterprise IT and cloud-native organizations have rapidly adopted AI-driven monitoring and predictive maintenance tools; this is now standard practice in information-sector infrastructure management and growing in mid-market deployments.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations and infrastructure management have moderate AIOps adoption with growing pilots, but many organizations still rely on manual review, especially outside large enterprises.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI systems significantly enhance administrator productivity by continuously surfacing anomalies, correlating multi-source data, and flagging maintenance needs without requiring manual log inspection, enabling humans to focus on remediation strategy rather than detection.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring and analytics tools substantially help administrators spot trends, anomalies, and degradation faster, meaningfully boosting productivity while humans retain final decision authority.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can reliably parse equipment logs, detect anomalies, and recommend maintenance or replacement based on performance metrics and failure patterns with minimal human intervention. Current systems achieve >50% time savings by automating trend analysis and threshold alerting that would otherwise require manual log review.
Task automatability	claude-sonnet-5	3/5	AI can analyze logs and performance metrics to flag anomalies and predict failures, but final repair/replace decisions often require contextual judgment about budgets, criticality, and vendor support that isn't fully automated end-to-end today. Roughly half the analytical workflow can be automated with monitoring/AIOps tooling.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While final replacement decisions often require human sign-off due to business impact and cost, there are minimal legal or regulatory barriers preventing automation of the analysis and recommendation phases. Most friction is organizational rather than structural.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement mandates a human perform this analysis, though organizational risk tolerance and internal change-management processes create some friction before acting on AI recommendations.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated monitoring and ML-based performance analysis cost a fraction of a system administrator's billable time for continuous equipment surveillance, particularly when considering 24/7 coverage and the scale at which these tools operate.
Cost vs. human wage	claude-sonnet-5	3/5	Monitoring platforms reduce manual log review time significantly, but licensing, integration, and tuning costs plus ongoing human oversight keep costs roughly comparable to dedicated admin time for smaller environments, though larger orgs see meaningful savings.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed monitoring and AIOps platforms (Datadog, New Relic, Splunk ML Toolkit) actively perform anomaly detection and predictive maintenance in production environments at scale, though interpretation still often requires human validation for final replacement decisions.
Technical feasibility today	claude-sonnet-5	3/5	AIOps and predictive maintenance products (e.g., SolarWinds, Datadog, ServiceNow) are deployed in production and do flag performance degradation and recommend actions, but they still have false positives/negatives and require human review before repair/replace decisions.

Monitor network performance to determine whether adjustments are needed and where changes will be needed in the future.

CI 57–75 · exposure 62 · 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	Information and technology sectors have rapidly adopted AI-powered network monitoring and observability tools. Major cloud providers, financial institutions, and large enterprises now routinely deploy ML-based monitoring systems, indicating deep and fast adoption in high-digitization sectors.
Sector adoption velocity	claude-sonnet-5	4/5	IT operations and infrastructure teams are among the faster adopters of AI-based monitoring and predictive analytics tools, reflecting the broader fast adoption pattern in technical/professional services.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments network administrators by automating routine monitoring, flagging anomalies in real time, and suggesting capacity planning changes, freeing humans to focus on strategic architecture and complex troubleshooting. The human remains essential for validation and judgment on major changes.
Augmentation potential	claude-sonnet-5	4/5	AI monitoring tools significantly boost administrators' ability to detect performance issues and forecast capacity needs, letting them focus attention on prioritized problems rather than manual log review.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can continuously monitor metrics, analyze network traffic patterns, and generate alerts or change recommendations with minimal human intervention. Modern observability platforms use ML for anomaly detection and threshold-based alerts, achieving significant time savings (50%+) on routine monitoring and many predictive recommendations, though some novel situations still require human judgment.
Task automatability	claude-sonnet-5	3/5	AI-driven network monitoring tools (AIOps) can flag anomalies and predict capacity issues, but interpreting complex infrastructure changes and deciding future architecture still requires human judgment and context.rationale removed
Adoption barriers	claude-haiku-4-5-20251001	2/5	Network monitoring decisions rarely require regulatory sign-off or licensing. Organizations may prefer humans in the loop for critical changes, but this is preference-based friction rather than a hard legal barrier. Integration into existing tooling and some stakeholder buy-in are the main obstacles.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement blocks automation of monitoring itself, though organizational risk aversion around critical infrastructure changes creates some friction before fully autonomous action is trusted.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven monitoring platforms cost a fraction of the hourly wage of dedicated network operations staff monitoring networks 24/7. A single AI system can monitor thousands of networks simultaneously, achieving an order-of-magnitude cost advantage over hiring humans for continuous real-time monitoring.
Cost vs. human wage	claude-sonnet-5	3/5	Monitoring software has licensing and infrastructure costs comparable to a portion of an administrator's time saved, but human oversight for interpretation and planning keeps costs roughly comparable rather than dramatically cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature products (e.g., Datadog, New Relic, Splunk, Cisco AI Network Analytics) reliably perform network monitoring, anomaly detection, and performance forecasting in production at scale. They demonstrably reduce on-call burden and catch issues early, though they still generate false positives and require human validation for some recommendations.
Technical feasibility today	claude-sonnet-5	3/5	AIOps and network monitoring platforms with anomaly detection and predictive alerts are deployed in production, but they still require significant human tuning and interpretation, and error rates on novel issues remain notable.

Train people in computer system use.

CI 41–87 · exposure 58 · augmentation 88 · 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	4/5	IT organizations have rapidly adopted LMS systems, AI tutoring, and self-service knowledge bases; major vendors (Microsoft, Cisco, others) now embed AI-assisted training into product suites. This reflects fast adoption in the digitally mature, information-sector domains where network admins operate.
Sector adoption velocity	claude-sonnet-5	3/5	IT departments increasingly use AI chatbots and self-service documentation for training, but structured live training programs still commonly involve human administrators.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI dramatically augments trainers by automating content creation, enabling personalized learning paths, handling routine Q&A, and freeing trainers to focus on mentorship and organizational-specific problem-solving. Human trainers using AI assistance can serve far larger populations while improving learning outcomes.
Augmentation potential	claude-sonnet-5	4/5	AI can generate training materials, FAQs, quizzes, and personalized learning paths, significantly boosting an administrator's efficiency in preparing and delivering training.
Task automatability	claude-haiku-4-5-20251001	5/5	AI can generate comprehensive training materials, create interactive tutorials, deliver instructional content, and answer system-use questions at scale with minimal human intervention. This task is well-suited to automation, as it involves content delivery and knowledge transfer that AI can execute end-to-end with >50% time savings compared to human trainers.
Task automatability	claude-sonnet-5	2/5	Training often requires live demonstration, adapting to trainee questions, hands-on troubleshooting, and reading the room, which current AI cannot fully replicate end-to-end though it can generate materials.rings
Adoption barriers	claude-haiku-4-5-20251001	2/5	While organizations often prefer human interaction for complex or organizational-specific training, no licensing requirement or strict legal mandate exists for training delivery itself. IT departments can and do substitute AI-driven training with minimal regulatory friction, though customer preference and organizational inertia provide modest friction.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement for internal IT training, but organizations often prefer human trainers for onboarding and complex troubleshooting, creating moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI training systems operate at near-zero marginal cost per trainee after initial setup, versus loaded cost of human trainers ($50–150k/year+benefits for significant audience coverage). The cost advantage easily exceeds an order of magnitude for large-scale deployment.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply produce training content and answer basic questions, but live instruction, Q&A, and customization still require human time, making costs roughly comparable for full training delivery.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed AI systems (LLMs, learning management system integrations, chatbots) already reliably handle training content delivery and Q&A at scale in production environments. Slight deduction because live training may still benefit from human engagement nuance, though pre-recorded and AI-assisted instruction is demonstrably effective.
Technical feasibility today	claude-sonnet-5	2/5	AI-generated tutorials, chatbots, and documentation exist but deployed products rarely deliver full interactive training sessions reliably without human facilitation.

Configure, monitor, and maintain email applications or virus protection software.

CI 53–75 · exposure 62 · augmentation 75 · 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	IT organizations, especially in mid-to-large enterprises and cloud-native settings, have widely adopted automation frameworks and managed services for email and endpoint protection, though smaller organizations and legacy environments lag.
Sector adoption velocity	claude-sonnet-5	4/5	IT and cybersecurity sectors show fast, deep adoption of AI-driven monitoring and antivirus/email security tools, with automated threat detection and patching now standard in most production environments.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-assisted alerting, anomaly detection, and automated remediation suggestions significantly enhance administrator productivity by reducing triage time and flagging threats faster than manual review alone.
Augmentation potential	claude-sonnet-5	4/5	AI significantly augments administrators by automating alerting, anomaly detection, and routine maintenance tasks, letting humans focus on complex configuration and incident response decisions.
Task automatability	claude-haiku-4-5-20251001	4/5	Configuration of email and virus protection software can be largely automated through infrastructure-as-code, policy templates, and monitoring systems; however, troubleshooting complex incidents and responding to novel threats typically requires human judgment, keeping it shy of full automation at equal quality.
Task automatability	claude-sonnet-5	3/5	AI can automate routine configuration templates, monitoring alerts, and signature updates, but troubleshooting novel issues, network-specific tuning, and incident response still require human judgment and hands-on system access.'
Adoption barriers	claude-haiku-4-5-20251001	3/5	Organizational and security policies typically require human sign-off on critical configurations, and internal compliance/audit processes create friction; however, no legal license requirement blocks automation of routine deployment and monitoring.
Adoption barriers	claude-sonnet-5	2/5	No strict licensing requirement dictates a human must perform this task, though organizational policies often require accountable staff for security-critical systems and compliance sign-off in regulated industries.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	Cloud-based and API-driven management tools significantly reduce per-instance administration costs compared to manual per-host configuration, and continuous monitoring through agents is orders of magnitude cheaper than human spot-checking.
Cost vs. human wage	claude-sonnet-5	3/5	Automated monitoring and security tools reduce labor hours, but licensing costs for enterprise-grade AI security software plus required human oversight keep costs roughly comparable to a skilled administrator's time for full-scope maintenance.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (e.g., Microsoft Intune, Jamf, Cisco Umbrella, SentinelOne) reliably automate deployment, patching, and routine monitoring in production environments; edge cases and custom policy tuning still require administrative oversight, but core functionality is mature and widely used.
Technical feasibility today	claude-sonnet-5	3/5	Products like AI-driven SIEM/EDR tools and email security platforms (e.g., Microsoft Defender, Proofpoint) reliably automate threat detection and filtering, but full lifecycle configuration and maintenance across heterogeneous environments still needs admin oversight.

Maintain an inventory of parts for emergency repairs.

CI 46–60 · exposure 42 · augmentation 75 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Many IT organizations use automated inventory and asset management tools, but full autonomy in emergency repair part allocation remains limited; most deployments involve human decision-making on top of AI-assisted tracking.
Sector adoption velocity	claude-sonnet-5	3/5	IT departments have moderate adoption of asset/inventory management software, but many organizations still rely on manual spreadsheets or ad hoc tracking, especially smaller firms.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-driven inventory systems significantly enhance human productivity by automating tracking, predicting stock needs, flagging low supplies, and optimizing part allocation—allowing administrators to focus on emergency response rather than manual stock counts.
Augmentation potential	claude-sonnet-5	4/5	AI-enabled inventory systems can predict shortages, automate reordering, and flag anomalies, significantly boosting efficiency for the human managing the process.
Task automatability	claude-haiku-4-5-20251001	2/5	Current AI can assist with inventory tracking and database updates, but the task requires physical stock management, vendor coordination, and contextual decision-making about part allocation during emergencies—elements that demand human judgment and physical presence.
Task automatability	claude-sonnet-5	3/5	Inventory tracking and reorder triggers can largely be automated with software systems, but physical verification, procurement decisions, and edge-case judgment still require human involvement.4
Adoption barriers	claude-haiku-4-5-20251001	2/5	Minimal legal barriers exist; inventory maintenance is not a licensed function. The main friction is organizational preference for human oversight during emergencies and the need for physical verification of stock, but these are not hard barriers to automation.
Adoption barriers	claude-sonnet-5	1/5	No licensing, regulatory, or liability barriers prevent automating inventory tracking for spare parts.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	Cloud-based inventory and asset management tools cost significantly less than a human managing inventory full-time, but the human must still verify physical stock and make allocation decisions, making the all-in cost roughly comparable to a part-time dedicated role.
Cost vs. human wage	claude-sonnet-5	3/5	Automated inventory systems reduce labor costs substantially, but licensing, integration, and periodic human verification keep costs from being an order of magnitude cheaper than human-managed processes.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Inventory management systems exist and can automate tracking and reordering, but they require human oversight for accuracy, physical verification, and emergency prioritization decisions; no fully autonomous system reliably manages emergency part allocation without human intervention.
Technical feasibility today	claude-sonnet-5	3/5	Inventory management software and IT asset management (ITAM) tools are mature and widely deployed, but full automation of emergency parts inventory still typically involves human oversight for accuracy and exceptions.

Operate master consoles to monitor the performance of computer systems and networks and to coordinate computer network access and use.

CI 41–55 · 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	3/5	Large enterprises and financial/tech sectors are deploying AI-assisted monitoring and anomaly detection actively, but small to mid-sized organizations lag significantly. Full autonomous operation remains rare; most deployments are in a hybrid human-AI oversight model rather than replacement.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations is a moderately fast-adopting sector for AIOps tools, though many organizations still rely heavily on human-monitored consoles, especially in smaller or legacy environments.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI dashboards, predictive alerting, and anomaly detection substantially augment human administrators by reducing noise, accelerating diagnostics, and freeing time for strategic work. The human remains the decision-maker on access and critical changes, but AI greatly amplifies their situational awareness and speed.
Augmentation potential	claude-sonnet-5	4/5	AI-driven dashboards, anomaly detection, and automated alerting significantly boost an administrator's ability to monitor systems and respond faster while retaining human control over access decisions.
Task automatability	claude-haiku-4-5-20251001	3/5	Current AI can monitor network metrics, detect anomalies, and generate alerts with reasonable accuracy, automating ~40-50% of the routine monitoring and initial triage. However, complex troubleshooting decisions, coordinating access across multiple systems with context-dependent rules, and handling edge cases still require human judgment, preventing full end-to-end automation.
Task automatability	claude-sonnet-5	3/5	AI-based monitoring tools can automate alerting, anomaly detection, and routine access coordination, but complex incident triage and judgment calls on network changes still require human oversight.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Access control and network security are heavily regulated (SOC 2, ISO 27001, HIPAA, PCI-DSS), and many organizations require a licensed or credentialed human to authorize and sign off on network access changes and critical system modifications. Liability and security breach consequences create strong organizational resistance to full automation.
Adoption barriers	claude-sonnet-5	2/5	No strict licensing barrier exists, but organizational policies often require human sign-off for network access changes and security-sensitive actions, creating moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	While monitoring AI has low inference costs, the infrastructure, integration, and mandatory human oversight to validate alerts and authorize access changes keep total cost comparable to or sometimes higher than a part-time network monitor's loaded wage.
Cost vs. human wage	claude-sonnet-5	3/5	Monitoring software licensing plus integration and oversight costs are often comparable to a portion of an administrator's salary, though it reduces the need for constant manual console watching.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Mature products like AI-driven network monitoring tools (e.g., Cisco's AI Network Analytics, Splunk ML Toolkit) perform anomaly detection and alerting reliably in production. However, complete master console operation with autonomous access coordination remains narrow in scope and often requires human validation before critical actions.
Technical feasibility today	claude-sonnet-5	3/5	AIOps and network monitoring platforms (e.g., Datadog, SolarWinds with AI features, ServiceNow) are deployed in production but typically augment rather than fully replace console operators, especially for access control decisions.

Diagnose, troubleshoot, and resolve hardware, software, or other network and system problems, and replace defective components when necessary.

CI 32–53 · 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	4/5	Large enterprises in information and financial sectors have rapidly deployed AIOps platforms and AI-assisted ticketing systems in production; smaller firms and physical infrastructure environments lag. Mid-to-high adoption velocity overall in digitized sectors, driven by cost and incident-response pressures.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations is a digitized sector adopting AIOps and monitoring tools at a moderate pace, with pilots common but full autonomous resolution still rare in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI assistants demonstrably improve admin productivity through automated log analysis, pattern suggestions, documentation search, and guided troubleshooting workflows while admins retain control and judgment. Many organizations report significant time savings in diagnosis phases through AI augmentation.
Augmentation potential	claude-sonnet-5	4/5	AI-powered monitoring, log analysis, and diagnostic assistants meaningfully speed up problem identification and root-cause analysis, letting administrators focus on complex fixes and physical repairs.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can assist with diagnosis through log analysis and pattern matching, and can guide troubleshooting workflows, but hardware replacement and complex multi-system issues requiring physical access or novel problem combinations still require human intervention. Roughly half the diagnostic and resolution workflow could be automated with significant setup.
Task automatability	claude-sonnet-5	2/5	AI can assist with log analysis and suggest diagnostic steps, but physical hardware replacement and complex, context-specific network troubleshooting still require hands-on human intervention and judgment.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Organizational friction and the requirement for human sign-off on critical systems and physical interventions create meaningful adoption friction, though no hard regulatory licensing requirement exists for the diagnostic portion. Customer preference for human technical staff and liability concerns for system downtime add friction.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement per se, but organizational risk aversion, need for physical access to hardware, and liability for critical system failures create meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI inference and integration costs for continuous monitoring and diagnosis are now comparable to the labor cost of first-tier support staff, but the need for human verification, escalation handling, and physical work maintains rough cost parity overall.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools can reduce time spent on initial triage, but the need for human oversight, physical intervention, and specialized troubleshooting keeps overall costs comparable to or only modestly cheaper than human labor.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products exist for automated monitoring, log analysis, and ticket triage (e.g., AIOps platforms, ChatGPT-based ticket analysis), but they have material error rates in root cause identification and cannot handle physical replacement or complex edge cases reliably. Deployment is common in larger organizations but with human oversight remaining essential.
Technical feasibility today	claude-sonnet-5	2/5	Some AIOps and monitoring tools exist that flag anomalies and suggest fixes, but reliable end-to-end diagnosis and resolution across diverse hardware/software stacks in production is still narrow and error-prone.

Confer with network users about solutions to existing system problems.

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	Many enterprises have piloted AI-assisted ticketing and chatbots for first-line support, but true end-to-end automation of user conferences remains limited; adoption is exploratory and mixed, with most organizations retaining humans in critical troubleshooting loops.
Sector adoption velocity	claude-sonnet-5	3/5	IT/tech sectors adopt AI support tools quickly, but full replacement of interactive troubleshooting conversations remains at pilot/production-assist stage.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist by suggesting solutions, drafting explanations, pulling historical ticket data, and auto-generating diagnostics in real time, meaningfully boosting the admin's productivity while the human manages the dialogue and makes final decisions.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly help by summarizing logs, suggesting fixes, and drafting explanations, greatly speeding up the human's ability to confer with users.
Task automatability	claude-haiku-4-5-20251001	2/5	Conferring with users requires nuanced dialogue, context-sensitive problem diagnosis, and interpersonal judgment. While AI can draft responses or suggest solutions from logs, the interactive back-and-forth, empathy, and real-time adaptation to user concerns remain largely manual; full end-to-end automation achieving 50% time savings at equal quality is not demonstrated at scale.
Task automatability	claude-sonnet-5	2/5	This requires live diagnostic conversation, contextual judgment about specific user environments, and often follow-up troubleshooting that current AI cannot fully replace end-to-end.'
Adoption barriers	claude-haiku-4-5-20251001	3/5	Organizations often prefer human contact for sensitive system problems, and liability concerns around incorrect advice create friction, but there is no strict legal barrier preventing AI assistance; organizational preference and error-cost asymmetry provide moderate protection.
Adoption barriers	claude-sonnet-5	2/5	No strict licensing requirement, but organizational trust, security sensitivity, and preference for human judgment in diagnosing system issues create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-powered support tools (chatbots, ticket systems) have modest infrastructure costs but require significant human oversight, escalation, and follow-up; the combined cost of inference, integration, and mandatory human review likely approaches or exceeds the marginal cost of a trained admin directly handling the call.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted triage lowers cost for simple queries, but complex conferring still requires human admin time, keeping overall cost roughly comparable for many cases.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably conducts user-facing troubleshooting calls or conferences end-to-end. Chatbots exist but struggle with domain complexity, context retention, and the nuance required to understand user descriptions of problems; production systems are narrow and require human escalation.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI helpdesk tools exist and handle simple tickets, but reliable, nuanced conferring about complex network problems in production is still narrow and error-prone.

Research new technologies by attending seminars, reading trade articles, or taking classes, and implement or recommend the implementation of new technologies.

CI 30–41 · exposure 25 · augmentation 75 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Most organizations retain humans for strategic technology decisions; while IT teams use AI for preliminary research aggregation, actual adoption of AI-driven technology recommendations remains limited and cautious in production settings.
Sector adoption velocity	claude-sonnet-5	3/5	IT/professional services sectors adopt AI research and summarization tools at a moderate pace, with pilots for AI-assisted knowledge work common but full replacement of technology evaluation processes still rare.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at helping administrators gather, filter, and summarize information from seminars, articles, and class materials, substantially reducing research time while humans retain decision-making authority on fit and implementation strategy.
Augmentation potential	claude-sonnet-5	4/5	AI significantly aids by summarizing trade publications, generating comparisons of technologies, and drafting recommendation reports, meaningfully speeding up the research portion of this task while humans retain decision authority.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can summarize trade articles and search seminar materials, the task requires human judgment to evaluate applicability to specific organizational contexts, assess fit with existing infrastructure, and make recommendations that account for strategic priorities—activities beyond current AI capabilities at scale.
Task automatability	claude-sonnet-5	2/5	AI can help surface and summarize new technologies (e.g., via search/synthesis), but attending seminars, evaluating vendor claims in context, and making implementation recommendations for a specific network environment require judgment and organizational knowledge AI cannot fully replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Implementation decisions often require sign-off from IT leadership and carry liability if they fail, creating organizational friction. However, research and recommendation roles are not strictly licensed, allowing some delegation to AI-assisted workflows with human approval.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational trust and accountability for technology decisions typically stay with human staff, creating moderate friction against full delegation to AI.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for research aggregation are inexpensive, but implementation recommendations require skilled human administrators whose judgment cannot yet be fully replaced; the all-in cost including human oversight remains comparable to or higher than direct human performance.
Cost vs. human wage	claude-sonnet-5	3/5	AI tools are cheap for literature/trend scanning, but the overall task still requires substantial human time for evaluation, vendor engagement, and contextual recommendation, keeping costs roughly comparable to human effort when done properly.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs end-to-end technology research, evaluation, and recommendation for organizational implementation. AI tools exist for content aggregation and summarization, but actual technology vetting and implementation decisions remain human-driven in production environments.
Technical feasibility today	claude-sonnet-5	2/5	Products like AI research assistants and summarization tools exist and are used informally, but no deployed product reliably performs the full research-to-recommendation cycle for IT infrastructure decisions in production.

Recommend changes to improve systems and network configurations, and determine hardware or software requirements related to such changes.

CI 25–42 · exposure 30 · 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	2/5	While IT organizations pilot AI-assisted diagnostics, production adoption of AI-driven system recommendations remains slow and cautious. Most enterprises rely on human administrators for final decisions, and the conservative posture toward infrastructure changes limits rapid AI deployment in this domain.
Sector adoption velocity	claude-sonnet-5	3/5	IT operations is a moderately fast-adopting sector with growing use of AIOps and monitoring tools, but full delegation of configuration recommendation authority to AI remains uncommon in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist administrators by analyzing logs, identifying performance bottlenecks, and generating candidate recommendations for review. Tools that surface data-driven insights while keeping the human in control substantially boost productivity without replacing professional judgment.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully assist by analyzing logs, flagging inefficiencies, drafting configuration options, and summarizing best practices, substantially speeding up the human's diagnostic and planning process.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze current configurations and suggest optimizations, recommending changes requires understanding organizational context, business constraints, and risk tolerance that typically demand human judgment. AI struggles with the integration of complex interdependencies and the accountability required for infrastructure decisions.
Task automatability	claude-sonnet-5	2/5	AI can suggest configuration improvements based on documented context, but this task requires deep knowledge of the specific network's history, business constraints, and undocumented quirks that current systems cannot fully access or reason about end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Network infrastructure decisions carry high liability and compliance requirements (security, uptime, regulatory mandates). Organizations typically require a human administrator's sign-off and expertise for system recommendations, and errors can be costly, creating both legal and operational accountability barriers to full automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement dictates a human must make these recommendations, but organizational risk aversion and the criticality of infrastructure decisions create moderate friction against fully trusting AI outputs.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for configuration analysis and recommendations exist but are often expensive specialized software requiring significant integration and oversight costs. A skilled administrator's cost is partially comparable, and the savings do not yet reach an order of magnitude difference when accounting for validation and liability.
Cost vs. human wage	claude-sonnet-5	2/5	Significant human oversight and validation is still needed to trust and implement recommendations, so AI reduces effort but does not yet approach an order-of-magnitude cost advantage over the sysadmin's fully-loaded time.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Tools like network analysis software and some AI-assisted diagnostic systems can suggest configuration improvements and identify hardware gaps in controlled environments, but production deployment remains limited. Current systems often produce false positives or miss nuanced organizational requirements, requiring significant human validation.
Technical feasibility today	claude-sonnet-5	2/5	AI-assisted network analysis tools and copilots exist but are narrow (e.g., anomaly detection, config linting); no deployed product autonomously produces reliable holistic recommendations for hardware/software changes across an org's real environment.

Gather data pertaining to customer needs, and use the information to identify, predict, interpret, and evaluate system and network requirements.

CI 30–38 · exposure 25 · augmentation 63 · 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	IT operations has moderate digitization but adoption of AI for requirements gathering is still limited to experimentation with analytics tools; most organizations rely on traditional methods (interviews, audits, ticketing system analysis) and have not shifted to AI-driven approaches in production.
Sector adoption velocity	claude-sonnet-5	3/5	IT and enterprise sectors show moderate AI adoption for network monitoring and diagnostics, but requirements-gathering with customers remains a human-centric, pilot-stage area at best.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by surfacing patterns in ticket data, predicting resource bottlenecks from historical logs, and drafting preliminary requirements summaries, which accelerates the administrator's analysis phase. However, the human must validate findings and conduct stakeholder interviews, so augmentation is real but bounded.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by analyzing usage data, summarizing customer input, drafting requirement specs, and modeling capacity needs, significantly speeding up the analytical portions of this task.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data collection and some pattern recognition in customer feedback, gathering nuanced customer needs requires contextual understanding and follow-up clarification that AI cannot reliably perform end-to-end. The interpretive and predictive components around system requirements demand domain judgment and stakeholder interaction that falls short of 50% time savings at equal quality.
Task automatability	claude-sonnet-5	2/5	This requires stakeholder interviews, contextual judgment about business needs, and synthesis with organizational knowledge that current AI cannot autonomously gather or validate end-to-end.assistance is possible for analysis but not full task substitution.
Adoption barriers	claude-haiku-4-5-20251001	3/5	System design and capacity planning typically require human sign-off and accountability, and customers often expect direct engagement with named IT professionals. Organizational practices favor human-led requirements gathering, though no formal legal barrier prevents AI assistance in this administrative task.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement specifically for this task, but organizational trust, stakeholder relationships, and accountability for infrastructure decisions create moderate friction against pure automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The integration costs of AI tools (data cleaning, validation, human oversight to correct misinterpretations) and the necessity of skilled human review of outputs keep total cost comparable to or higher than direct human effort. The task's complexity and stakes make automation economically marginal today.
Cost vs. human wage	claude-sonnet-5	2/5	Human elicitation of needs via conversation and relationship-building remains cheaper and more reliable than AI-driven equivalents once integration and oversight costs are factored in.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably perform this full task in production; existing tools help analyze logs or extract structured data from surveys, but the synthesis of customer needs into validated system requirements still requires human administrators. Diagnostic and advisory tools exist but lack the collaborative discovery element of this task.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously gathers customer requirements and translates them into network specifications reliably; existing tools assist with documentation and analysis but don't perform the discovery process itself.

Coordinate with vendors and with company personnel to facilitate purchases.

CI 25–35 · exposure 25 · augmentation 50 · 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	2/5	IT procurement workflows remain largely human-directed and conservative; while pilot AI assistants for vendor RFQ drafting exist, deep production automation of purchase coordination across IT departments remains uncommon and slow to scale.
Sector adoption velocity	claude-sonnet-5	2/5	IT procurement processes in most organizations remain manually driven with limited AI-driven vendor coordination in production, especially for mid-sized firms typical of this role.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist by summarizing vendor proposals, drafting emails, organizing cost comparisons, and flagging duplicate requests—activities that reduce clerical overhead while the administrator retains decision authority and vendor relationships.
Augmentation potential	claude-sonnet-5	3/5	AI can help draft RFPs, summarize vendor proposals, track communications, and organize purchase documentation, improving efficiency while humans retain decision-making.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help draft communications and organize vendor information, the task fundamentally requires negotiation, relationship management, and judgment about vendor suitability that depend on human discretion and accountability. Current systems cannot reliably coordinate multi-party procurement decisions end-to-end with the quality and authority needed for IT purchasing.
Task automatability	claude-sonnet-5	2/5	This involves negotiation, relationship management, and cross-checking internal needs against vendor offerings, which requires judgment and interpersonal coordination beyond current AI capabilities to fully automate.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Procurement decisions typically require human authorization and signature for legal/budget accountability; company policy often mandates that designated personnel approve vendor selections. Liability and contract risk create high barriers to autonomous coordination.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement, but organizational approval chains, vendor relationships, and procurement policies create moderate friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tooling costs for purchase coordination (LLM API calls, integration, oversight by the administrator) approach or exceed the time savings, since these tasks already involve low per-unit overhead and much of the work is dialogue and decision-making rather than rote process.
Cost vs. human wage	claude-sonnet-5	2/5	Human coordination still requires oversight and relationship-building that AI cannot fully replace, so all-in AI costs including human review are not dramatically cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably orchestrates vendor coordination and company-wide procurement approval workflows autonomously. Tools exist for email drafting and vendor database management, but they function as narrow helpers rather than systems that can independently facilitate purchases across organizational contexts.
Technical feasibility today	claude-sonnet-5	2/5	AI tools can draft communications or summarize vendor quotes, but no deployed product autonomously manages the full vendor coordination and purchasing facilitation process reliably.

Maintain and administer computer networks and related computing environments, including computer hardware, systems software, applications software, and all configurations.

CI 28–32 · exposure 25 · 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	3/5	Large enterprises and financial institutions have deployed limited AI-assisted monitoring and orchestration, but adoption remains primarily in pilot and early production phases; small and mid-market organizations lag significantly.
Sector adoption velocity	claude-sonnet-5	3/5	IT/tech sector has moderate-to-fast adoption of automation and AIOps tools, with many pilots and growing production use of monitoring/remediation bots, though full network administration automation remains uncommon.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI systems demonstrably assist administrators through automated alerting, anomaly detection, patch recommendation, and log aggregation, meaningfully raising productivity and reducing manual busywork while humans retain decision-making authority.
Augmentation potential	claude-sonnet-5	4/5	AI-driven monitoring, anomaly detection, automated ticketing, and config-management copilots meaningfully boost administrator productivity across large parts of this task while humans retain oversight and handle hardware/complex issues.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with routine configuration management and basic monitoring tasks, the full scope—hardware maintenance, software updates across diverse systems, troubleshooting complex failures, and handling edge cases—requires human judgment, domain expertise, and physical intervention that AI cannot yet replace end-to-end with 50% time savings at equal quality.
Task automatability	claude-sonnet-5	2/5	This is a broad, ongoing operational task spanning hardware, software, and configuration management that requires physical access, judgment, and troubleshooting across heterogeneous systems, limiting end-to-end automation today. Sub-tasks like monitoring and scripted config changes are automatable, but the full scope is not.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory compliance (SOX, HIPAA, PCI-DSS), liability for security breaches or downtime, and organizational requirements for human accountability and sign-off on critical infrastructure changes create strong legal and procedural barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	No formal licensing requirement, but high liability for network outages/security breaches, need for physical hardware access, and organizational risk-aversion around critical infrastructure create meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI monitoring and automation tools require integration, licensing, and continuous human oversight by skilled administrators, meaning total cost approaches or exceeds the loaded wage of mid-tier network staff rather than achieving meaningful cost savings.
Cost vs. human wage	claude-sonnet-5	2/5	Automation tools reduce some routine costs, but licensing, integration, and the need for skilled oversight of critical infrastructure keep AI-assisted administration cost-comparable rather than dramatically cheaper than human admins for the full task scope.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Current AI products exist for narrow aspects (alert filtering, log analysis) but no production system reliably handles the full spectrum of network administration autonomously; most deployments require substantial human oversight and fallback for critical decisions.
Technical feasibility today	claude-sonnet-5	2/5	AIOps and infrastructure-as-code tools exist and are used for monitoring, alerting, and automated remediation of known issues, but comprehensive administration including hardware and novel troubleshooting still relies heavily on human admins in production environments.

Design, configure, and test computer hardware, networking software and operating system software.

CI 25–32 · 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	2/5	Adoption in this domain remains slow relative to lower-stakes IT work; most organizations still rely on human teams for design and formal testing, with AI playing only an assistive role in script generation or documentation. Full automation in production settings remains rare.
Sector adoption velocity	claude-sonnet-5	3/5	IT/tech sector adopts AI tools moderately fast for scripting and troubleshooting assistance, but full design/config/test automation pilots are less mature than in pure software domains.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments human administrators by auto-generating configuration templates, drafting test cases, suggesting optimization patterns, and documenting designs, enabling experienced engineers to design and validate systems faster while retaining full control and judgment.
Augmentation potential	claude-sonnet-5	4/5	AI significantly helps with generating configuration scripts, troubleshooting, documentation, and test case generation, meaningfully boosting admin productivity while humans remain in control.
Task automatability	claude-haiku-4-5-20251001	2/5	Design and testing of complex systems require significant human judgment about requirements, tradeoffs, and validation. While AI can assist with configuration templating and routine testing scripts, end-to-end design and comprehensive testing with equal quality remain beyond current AI capability without substantial human oversight.
Task automatability	claude-sonnet-5	2/5	Design and testing of hardware/network architectures require contextual judgment, physical setup, and integration testing that current AI can assist but not fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Significant liability, regulatory, and organizational barriers exist: system failures can cause major outages, many enterprises require certified professionals to sign off on designs, compliance frameworks (SOC 2, ISO 27001, etc.) often mandate human architect review and governance, and enterprise risk aversion limits substitution.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement, but security, compliance, and liability concerns around misconfigured networks create meaningful organizational caution against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI coding and configuration assistants reduce some design and scripting work, but the loaded cost of a skilled network/systems administrator remains lower than the all-in cost (inference, integration, specialized oversight, and liability) for AI to produce production-grade designs and test coverage.
Cost vs. human wage	claude-sonnet-5	2/5	Human sysadmins still need to validate and physically implement configurations, so AI reduces some effort but doesn't eliminate the labor cost, keeping costs roughly comparable to somewhat favorable.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs complete design-configure-test cycles autonomously; tools exist for narrower tasks like config generation or automated testing of specific components, but production systems require human architects and QA engineers to direct the process and validate results.
Technical feasibility today	claude-sonnet-5	2/5	AI tools (config generators, network design assistants) exist but are narrow-scope aids; no deployed product autonomously designs, configures, and tests full network/OS stacks in production.

Load computer tapes and disks, and install software and printer paper or forms.

CI 21–35 · exposure 20 · augmentation 13 · importance 3.4/5 · 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 cloud and virtualization have reduced tape/disk loading in many organizations, legacy systems in finance, healthcare, and government still require hands-on media management, limiting adoption velocity of full automation.
Sector adoption velocity	claude-sonnet-5	2/5	IT operations increasingly virtualize infrastructure, reducing physical media tasks, but the underlying physical task itself sees little direct AI adoption since it requires manual intervention.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI can assist with scheduling and inventory tracking of media, but the core physical loading task does not benefit meaningfully from AI assistance while a human remains present.
Augmentation potential	claude-sonnet-5	1/5	AI offers essentially no assistance for the physical act of loading tapes, disks, or paper, as this is a manual, non-cognitive task.
Task automatability	claude-haiku-4-5-20251001	2/5	Loading tapes and disks is partially automatable with existing tape libraries and disk management systems, but installing software and printer paper/forms involves physical manipulation that current robots cannot reliably handle at scale, and integration with legacy systems is complex.
Task automatability	claude-sonnet-5	2/5	This is a physical, hands-on task involving handling hardware media and consumables; AI software cannot physically load tapes/disks or paper without robotic embodiment, which is not standard today.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Physical infrastructure compatibility and legacy system requirements create moderate friction; however, no licensing or legal requirement prevents automation of these mechanical tasks.
Adoption barriers	claude-sonnet-5	2/5	No licensing or legal barriers exist, but the physical nature of accessing hardware, secure server rooms, and consumables creates practical friction against remote automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Tape loaders and automated disk management exist but carry high capital costs; the ongoing labor savings do not yet achieve cost parity with wages when accounting for specialized equipment and maintenance.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI-based substitute performing this physical task, so any 'AI cost' comparison is moot—human labor remains the only viable and cheaper option.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some aspects like software installation are automated in modern environments, but physical tasks like loading media and paper remain predominantly manual; mature products exist for software deployment but not for general hardware loading.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI products physically perform tape loading, disk insertion, or paper loading in production; this remains a manual IT task performed by humans.

Plan, coordinate, and implement network security measures to protect data, software, and hardware.

CI 28–28 · exposure 25 · augmentation 75 · 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	3/5	IT and finance sectors are adopting security automation (SOAR platforms, automated compliance) at moderate pace, but fundamental security architecture planning remains human-led. Pilots of autonomous security tools exist, but production-grade autonomous security planning adoption is still limited.
Sector adoption velocity	claude-sonnet-5	3/5	IT and cybersecurity is a digitized, fast-moving sector adopting AI-assisted security tools (anomaly detection, automated patching suggestions), but full autonomous security administration remains uncommon in production.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI assists significantly with threat detection, vulnerability identification, compliance monitoring, and incident analysis, helping administrators work faster and catch threats earlier. These tools substantially augment human security analysts' productivity while keeping humans accountable for strategic decisions and policy enforcement.
Augmentation potential	claude-sonnet-5	4/5	AI significantly augments this task via automated threat detection, log analysis, vulnerability prioritization, and policy drafting, letting administrators focus on judgment calls and implementation decisions.
Task automatability	claude-haiku-4-5-20251001	2/5	Network security planning requires human judgment on risk assessment, organizational context, and strategic decisions. While AI can assist with vulnerability scanning and routine implementation tasks, the coordination and decision-making phases—assessing organizational threat models, choosing appropriate measures, and validating security posture—still require human oversight and expertise.
Task automatability	claude-sonnet-5	2/5	Security planning and coordination require contextual judgment about business risk, existing infrastructure, and compliance needs that current AI cannot autonomously determine, though AI can assist with specific sub-tasks like config generation or vulnerability scanning.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Network security planning and implementation often requires licensed certifications (CISSP, Security+) and organizational accountability for breach prevention. Legal liability for security failures, regulatory compliance mandates (SOC 2, ISO 27001), and the critical nature of security decisions create strong barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Network security often falls under regulatory/compliance frameworks (HIPAA, PCI-DSS, SOC2) requiring accountable human sign-off, and liability for breaches strongly incentivizes human oversight and authorization structures.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for security (SIEM, vulnerability scanners) require significant integration, configuration, and oversight labor. The human expertise cost for security planning and implementation remains high because errors are costly; AI does not yet substantially reduce the total cost of secure network deployment.
Cost vs. human wage	claude-sonnet-5	2/5	Security tooling with AI features reduces some labor but still requires skilled administrators to interpret findings, configure systems, and maintain accountability, so total cost savings are moderate rather than order-of-magnitude.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature product autonomously handles the full planning, coordination, and implementation of network security measures end-to-end. Security tools exist for detection and response, but deploying a security architecture requires human security architects and administrators to make critical trade-off decisions that current AI systems cannot reliably perform alone.
Technical feasibility today	claude-sonnet-5	2/5	AI-driven security tools (SIEM with ML, automated vulnerability scanners) exist in production, but full planning, coordination, and implementation of a security architecture is still human-led with AI as a supporting tool.

Implement and provide technical support for voice services and equipment, such as private branch exchange, voice mail system, and telecom system.

CI 25–30 · exposure 25 · augmentation 50 · 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	2/5	Voice system administration is typically found in enterprise IT departments that have been slow to adopt AI-driven automation for core infrastructure; adoption remains limited to pilot chatbots for help desk functions rather than production system management or implementation.
Sector adoption velocity	claude-sonnet-5	2/5	IT infrastructure/telecom support in smaller organizations tends toward slower digitization; AI adoption here lags behind software-centric IT tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist with diagnostic recommendations, documentation retrieval, knowledge base search, and troubleshooting workflows, raising efficiency for human administrators. However, the assistance is primarily informational rather than transformative, as implementation and support remain largely human-driven.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with documentation, configuration scripts, log analysis, and troubleshooting guidance, improving efficiency while a human still performs physical and system-level tasks.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with basic troubleshooting and documentation of voice systems, the task involves hands-on equipment configuration, physical deployment, and system integration that requires direct technical intervention and contextual decision-making. Current AI cannot reliably handle the full scope end-to-end with 50% time savings.
Task automatability	claude-sonnet-5	2/5	Physical installation, cabling, and hands-on hardware troubleshooting for PBX and telecom equipment cannot be done by AI; only diagnostic/config-scripting portions are automatable today.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Telecom system implementation and support often requires certified technicians, vendor-specific licensing, and regulatory compliance (FCC rules, telecom regulations). Legal liability for service disruptions and the need for human sign-off on critical infrastructure changes create strong barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement typically, but organizational reliance on physical presence, vendor support contracts, and hands-on equipment access create real friction against remote AI substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI assistance for voice system support remains narrower in scope than human expertise, and integration costs for monitoring and support automation are still material relative to specialist wages. Full replacement is not cost-effective given the complexity and liability.
Cost vs. human wage	claude-sonnet-5	2/5	Physical equipment work and on-site support still require paid human labor; AI tools reduce some diagnostic time but don't eliminate the core service cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably perform complete voice system implementation and support independently. AI can support diagnostics and help desk tasks but cannot configure complex PBX systems, manage telecom infrastructure, or provide full technical support without human oversight and intervention.
Technical feasibility today	claude-sonnet-5	2/5	Some AI-assisted network diagnostic and chatbot support tools exist for VoIP troubleshooting, but end-to-end implementation and hardware support still require human technicians on-site.

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.