Software Quality Assurance Analysts and Testers

15-1253.00
Median wage $104,300/yr186,740 employed (US)Rank #45 of 923 scored · top 5% by substitution

Develop and execute software tests to identify software problems and their causes. Test system modifications to prepare for implementation. Document software and application defects using a bug tracking system and report defects to software or web developers. Create and maintain databases of known defects. May participate in software design reviews to provide input on functional requirements, operational characteristics, product designs, and schedules.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution58
Exposure52
Augmentation81

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

30 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

30%

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%53

panel mean rating 3.1/5 → substitution pressure 53/100

Technical feasibility todayw 20%51

panel mean rating 3.1/5 → substitution pressure 51/100

Cost vs. human wagew 15%57

panel mean rating 3.3/5 → substitution pressure 57/100

Adoption barriersw 20%inverted — strong barriers lower the score72

panel mean rating 2.1/5 (barrier strength) → substitution pressure 72/100

Sector adoption velocityw 10%65

panel mean rating 3.6/5 → substitution pressure 65/100

Task breakdown (30 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.

Conduct historical analyses of test results.

84

CI 70–97 · exposure 83 · augmentation 100 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA are highly digitized sectors with strong adoption of analytics and automation tools; many organizations are already using dashboards and automated reporting for test result analysis.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are high-digitization, fast-adopting sectors where test analytics and AI-assisted DevOps tooling are already widely deployed.
Augmentation potentialclaude-haiku-4-5-202510015/5AI significantly augments analyst productivity by rapidly surfacing patterns, generating hypothesis-driven insights, and producing summaries that human testers can review and interpret, enabling faster decision-making and deeper investigation.
Augmentation potentialclaude-sonnet-55/5AI tools significantly speed up identification of flaky tests, regression patterns, and historical trend summarization, greatly boosting analyst productivity while humans still validate conclusions.
Task automatabilityclaude-haiku-4-5-202510015/5AI can comprehensively analyze test result data, identify patterns, trends, and anomalies across historical datasets, and generate summaries and insights with minimal human intervention, easily exceeding 50% time savings at equal or better quality.
Task automatabilityclaude-sonnet-54/5Analyzing historical test data to find trends, flaky tests, or regression patterns is a data-processing task well suited to AI, especially with structured test logs and dashboards.; most of the analytical work can be automated with existing tools plus LLM-based summarization.
Adoption barriersclaude-haiku-4-5-202510012/5Data access and governance policies within organizations may require oversight, and some teams may prefer human validation of conclusions, but there are no hard licensing or legal barriers preventing AI from conducting historical test result analysis.
Adoption barriersclaude-sonnet-51/5No licensing or regulatory requirement; this is an internal engineering analytics task with no human sign-off mandate.
Cost vs. human wageclaude-haiku-4-5-202510015/5Inference costs for AI-driven analysis are negligible compared to the loaded wage of a QA analyst performing manual historical review and report generation, easily achieving an order-of-magnitude cost advantage.
Cost vs. human wageclaude-sonnet-54/5Automated log parsing and trend analysis via scripts/AI is far cheaper than manual review of historical test data once pipelines are set up.
Technical feasibility todayclaude-haiku-4-5-202510015/5Deployed products (data analytics platforms, BI tools, LLMs with structured data analysis) reliably perform historical data analysis, trend identification, and report generation in production environments across many organizations today.
Technical feasibility todayclaude-sonnet-53/5Test analytics platforms (e.g., CI/CD dashboards, test intelligence tools) provide automated trend analysis, but deep historical root-cause synthesis still often requires human interpretation and customization per codebase.

Create or maintain databases of known test defects.

80

CI 67–92 · exposure 83 · augmentation 100 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA are information-sector roles with high digital maturity and rapid AI adoption; defect-tracking automation is already integrated into mainstream development platforms like Jira and GitHub, seeing fast and widespread deployment across tech companies.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA teams are moderately fast adopters of AI tooling, with pilots and partial production use of AI-assisted bug tracking common but not yet universal.
Augmentation potentialclaude-haiku-4-5-202510015/5AI significantly augments QA analysts' productivity by automating routine categorization, deduplication, and report ingestion while QA humans focus on triage logic, priority judgment, and investigating root causes, creating a strong human-AI partnership that multiplies analyst effectiveness.
Augmentation potentialclaude-sonnet-55/5AI tools strongly assist QA analysts by auto-classifying, summarizing, and flagging duplicate defects, significantly speeding up database maintenance while humans retain final oversight.
Task automatabilityclaude-haiku-4-5-202510015/5Creating and maintaining defect databases is a highly structured data task involving ingestion, classification, deduplication, and record management. Current AI systems can autonomously parse defect reports, extract relevant fields, detect duplicates, assign categories, and update database records with minimal human intervention, easily exceeding 50% time savings.
Task automatabilityclaude-sonnet-54/5Logging, categorizing, deduplicating, and updating defect records is a structured data-entry and pattern-matching task that current AI (especially LLM agents integrated with issue trackers like Jira) can largely automate, though some triage judgment remains human.
Adoption barriersclaude-haiku-4-5-202510012/5Defect database maintenance has minimal regulatory or licensing barriers; it is internal data management with no legal requirement for human sign-off, though some organizations may prefer human review for quality assurance before final entry into production systems.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates human maintenance of defect databases; main friction is organizational trust in AI-generated categorization and existing tool workflows.
Cost vs. human wageclaude-haiku-4-5-202510015/5The cost of inference plus API calls for database automation and deduplication is negligible compared to the loaded wage of a QA analyst manually logging, categorizing, and searching defect records, making AI cost an order of magnitude cheaper for equivalent output.
Cost vs. human wageclaude-sonnet-54/5Automating defect logging and database maintenance via AI tooling is far cheaper than manual QA analyst time spent on repetitive documentation, though integration and oversight costs are non-trivial.
Technical feasibility todayclaude-haiku-4-5-202510015/5Multiple deployed products reliably perform defect database management at scale today, including Jira, Azure DevOps, and specialized defect-tracking platforms with integrated AI features for categorization and deduplication that are in widespread production use across software organizations.
Technical feasibility todayclaude-sonnet-53/5Products exist (AI-assisted bug triage, Jira/GitHub Copilot integrations, auto-tagging tools) but most deployments still require human review for accuracy and prioritization, so reliability at scale is mixed.

Document software defects, using a bug tracking system, and report defects to software developers.

75

CI 70–80 · exposure 70 · augmentation 100 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA are highly digitized sectors with rapid AI adoption; continuous integration/continuous deployment (CI/CD) pipelines now routinely embed automated testing and defect logging. Large tech, finance, and enterprise software organizations are widely deploying these systems in production.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA is a fast-adopting sector for AI tooling, with many companies integrating AI copilots and bug-triage assistants into their pipelines already.
Augmentation potentialclaude-haiku-4-5-202510015/5AI-powered testing tools, dashboards, and automated defect reporting dramatically amplify QA analyst productivity by handling routine log entry, pattern detection, and prioritization, freeing analysts to focus on complex edge cases, root-cause analysis, and developer collaboration.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up drafting clear, well-structured defect reports from raw logs/screenshots while testers retain judgment on prioritization and validation.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automatically detect many software defects through testing, log them with structured data, and populate bug tracking systems with high accuracy. However, human judgment remains valuable for prioritization, severity classification, and developer communication nuance, limiting full end-to-end replacement but still achieving >50% time savings.
Task automatabilityclaude-sonnet-54/5Documenting defects with reproduction steps, severity, and logs into a bug tracker is a structured, language-heavy task that AI can largely draft or automate given test output and error logs, though some judgment on severity/triage remains.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or licensing barriers exist; bug tracking is primarily an operational and organizational practice. The main friction is organizational inertia around QA workflows and developer preference for human context-setting, not legal or compliance mandates.
Adoption barriersclaude-sonnet-51/5No licensing, legal, or human-contact requirement blocks automation of bug documentation; it's an internal engineering workflow task.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated defect detection and bug logging cost a small fraction of human QA analyst wages per defect instance, especially at scale. Infrastructure and maintenance add overhead, but the per-task cost is substantially lower than a human QA analyst's loaded compensation.
Cost vs. human wageclaude-sonnet-54/5Generating structured bug reports from logs/test failures via LLMs is cheap computationally compared to a QA engineer's time spent writing detailed reports manually.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature AI-assisted tools and automated testing platforms (e.g., Selenium-based systems, defect prediction models) are deployed in production across major software organizations. They reliably capture, categorize, and log defects at scale, though some edge cases and false positives still require human triage.
Technical feasibility todayclaude-sonnet-53/5AI-assisted bug report generation and auto-filing tools (e.g., integrated with CI/CD, Jira/GitHub plugins, LLM-based log summarizers) exist and are used in some orgs, but reliable end-to-end autonomous defect documentation is not yet universal or fully trusted.

Conduct software compatibility tests with programs, hardware, operating systems, or network environments.

74

CI 57–91 · exposure 67 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Software development sectors show very high adoption of CI/CD and automated testing: DevOps practices, GitHub Actions, and cloud testing platforms are now standard across startups, mid-market, and enterprise software shops.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA in tech/professional services sectors have rapidly adopted automated testing frameworks and AI-assisted test generation, though compatibility-specific automation lags full unit/functional test automation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-assisted test generation, anomaly detection in test results, and intelligent test prioritization meaningfully augment human QA engineers' productivity, allowing them to focus on complex scenarios and interpretation while automation handles routine coverage.
Augmentation potentialclaude-sonnet-54/5AI tools significantly speed up generation of test cases, scripts, and environment configurations, and can flag likely compatibility issues, greatly boosting tester productivity while humans validate and interpret results.
Task automatabilityclaude-haiku-4-5-202510014/5Compatibility testing is highly automatable: test harnesses, CI/CD pipelines, and automated test frameworks can execute predefined test suites across multiple OS/hardware/network configurations at scale, achieving >50% time savings. Human effort is mainly in test case design and interpretation of edge cases rather than test execution itself.
Task automatabilityclaude-sonnet-53/5AI can generate and execute compatibility test scripts and interpret results across configurations, but orchestrating diverse hardware/OS/network matrices and diagnosing subtle interaction failures still needs human setup and judgment., so only partial automation meets the 50% bar.
Adoption barriersclaude-haiku-4-5-202510012/5Minimal legal or licensing barriers exist; organizations can freely adopt automated testing. The main friction is internal: test case maintenance, integrating into workflows, and initial setup overhead—but these are operational, not regulatory.
Adoption barriersclaude-sonnet-51/5No licensing or legal requirement mandates human sign-off for compatibility testing; it's a purely technical QA function with no regulatory barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated testing infrastructure costs (cloud VMs, test frameworks) are orders of magnitude cheaper than human QA labor for equivalent test coverage, especially when amortized across hundreds of test cycles.
Cost vs. human wageclaude-sonnet-53/5Automated test execution is cheap per run, but building and maintaining compatibility test environments, device farms, and interpreting failures still requires significant skilled labor, keeping overall cost roughly comparable to human-run testing at scale.
Technical feasibility todayclaude-haiku-4-5-202510015/5Mature, production-deployed tools (Selenium, Jenkins, TestRail, BrowserStack, cloud-based CI/CD systems) reliably perform automated compatibility testing across diverse environments at scale across thousands of organizations.
Technical feasibility todayclaude-sonnet-53/5CI/CD pipelines with automated cross-browser/OS testing tools (e.g., Selenium grids, BrowserStack automation) are in production, but full compatibility testing across hardware/network permutations still requires human-configured test matrices and manual verification of edge cases.

Document test procedures to ensure replicability and compliance with standards.

72

CI 70–75 · exposure 70 · augmentation 100 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software and IT organizations are rapidly adopting AI coding and documentation tools in production; this is a core use case for agent-assisted development workflows and CI/CD automation pipelines.
Sector adoption velocityclaude-sonnet-54/5Software development and QA is a fast-adopting, highly digitized sector where AI coding/documentation assistants (e.g., Copilot-style tools) are already used in production pipelines at meaningful scale.
Augmentation potentialclaude-haiku-4-5-202510015/5AI excels at augmenting QA analysts by drafting templates, auto-generating procedure steps from code analysis, and flagging compliance gaps, allowing humans to focus on logic validation and edge cases rather than manual documentation labor.
Augmentation potentialclaude-sonnet-55/5AI tools substantially speed up drafting, formatting, and standardizing test documentation while the QA analyst remains responsible for accuracy, compliance verification, and final approval.
Task automatabilityclaude-haiku-4-5-202510014/5AI can substantially automate the generation, formatting, and organization of test procedure documentation by analyzing existing test cases, standards (ISO, NIST, etc.), and code. While human review of compliance accuracy and edge cases remains necessary, the time savings from automated drafting and structural compliance checking easily exceed 50%.
Task automatabilityclaude-sonnet-54/5Drafting structured test documentation from existing test cases, code, or requirements is largely a text-generation and summarization task that LLMs handle well, though final review against exact compliance standards still needs human sign-off., saving significant time.
Adoption barriersclaude-haiku-4-5-202510012/5While documentation must meet compliance standards and organizations often require human sign-off, there is no hard legal requirement that a licensed professional author test procedures. Organizational friction and quality oversight practices create some friction, but substitution is technically and legally feasible.
Adoption barriersclaude-sonnet-52/5No licensing requirement blocks AI-assisted documentation, though regulated industries (medical, aerospace, financial software) may require human-certified sign-off on compliance documentation, creating moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference cost for generating documentation is negligible compared to the labor cost of a QA analyst manually writing comprehensive test procedures; automation is roughly 10–20× cheaper per document when accounting for endpoint-to-endpoint workflow.
Cost vs. human wageclaude-sonnet-54/5Generating and formatting documentation via LLMs is far cheaper per unit output than a QA analyst manually writing detailed procedure docs, though some human review cost remains.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple deployed products (GitHub Copilot, Claude, specialized testing tools) reliably generate structured test documentation from test scripts and requirements with high accuracy. Production use is common in software teams, though some organizations still require human review before finalization due to liability concerns.
Technical feasibility todayclaude-sonnet-53/5AI coding assistants and QA tools can generate test case documentation and templates today, but reliably ensuring full compliance with specific organizational or regulatory standards across varied projects still shows material error rates in production use.

Monitor program performance to ensure efficient and problem-free operations.

72

CI 66–77 · exposure 67 · augmentation 100 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Software and IT operations sectors have rapidly adopted AI-driven monitoring and observability platforms; major cloud providers and DevOps organizations routinely deploy ML-based anomaly detection and automated alerting at scale.
Sector adoption velocityclaude-sonnet-54/5Software/tech sector has aggressively adopted AI-driven observability and monitoring platforms as standard practice, reflecting fast adoption patterns typical of information industries.
Augmentation potentialclaude-haiku-4-5-202510015/5AI monitoring systems substantially augment QA analysts by filtering noise, surfacing hidden patterns in performance data, and providing real-time alerts, allowing analysts to focus on investigation and resolution rather than manual log scanning.
Augmentation potentialclaude-sonnet-55/5AI monitoring tools significantly enhance testers' and engineers' ability to detect, triage, and prioritize performance issues, greatly boosting productivity while humans retain decision-making control.
Task automatabilityclaude-haiku-4-5-202510014/5AI can monitor program performance through logs, metrics, and automated alerts with high efficiency, capturing most anomalies and performance degradations. However, full end-to-end automation requires human judgment to contextualize alerts, triage false positives, and determine root causes, limiting it below a 5.
Task automatabilityclaude-sonnet-53/5AI-based monitoring tools can automatically detect performance anomalies and flag issues, but interpreting root causes and deciding remediation still requires human judgment, so only partial time savings are realized end-to-end.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or legal barriers exist to automating performance monitoring itself; organizations freely deploy ML-based APM and alerting systems. However, some friction arises from need for human validation of alerts and organizational preference for analyst oversight of critical systems.
Adoption barriersclaude-sonnet-52/5No licensing requirement blocks automation; some organizational friction exists around trusting automated alerts for critical production systems, but adoption is largely unrestricted.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven monitoring tools are significantly cheaper than hiring QA analysts to manually watch logs and metrics 24/7; the cost per monitored system-instance is substantially lower than human labor, though integration and alert tuning require upfront investment.
Cost vs. human wageclaude-sonnet-54/5Automated monitoring tools continuously watch systems at a fraction of the cost of manual monitoring by humans, though some engineer oversight and incident response remain necessary.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (APM tools with ML-driven anomaly detection, log analysis platforms) perform performance monitoring reliably in production at scale, though they still require human oversight to validate findings and respond to complex issues.
Technical feasibility todayclaude-sonnet-54/5APM products (Datadog, New Relic, Dynatrace) with AI-driven anomaly detection are widely deployed in production and reliably surface performance issues at scale today.

Identify program deviance from standards, and suggest modifications to ensure compliance.

71

CI 61–81 · exposure 62 · augmentation 100 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Adoption is rapid and deep across software development: automated code quality and compliance tools are ubiquitous in enterprise CI/CD pipelines, startups, and open-source projects, making this one of the most widely automated QA functions today.
Sector adoption velocityclaude-sonnet-54/5Software development and QA are high-digitization sectors with fast, deep adoption of AI-assisted code review and static analysis tools in production pipelines.
Augmentation potentialclaude-haiku-4-5-202510015/5AI-powered code analysis tools significantly augment QA analysts by automating routine deviation detection, freeing humans to focus on complex edge cases and architectural compliance decisions, thereby raising overall productivity while keeping humans in the loop.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up identification of standard deviations and drafts suggested fixes, letting QA analysts focus on validation and edge cases while remaining in the loop.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems can reliably identify deviations from coding standards, style guides, and documented compliance requirements through static analysis and pattern matching. Suggesting modifications to achieve compliance is automatable for well-defined standards (e.g., WCAG, OWASP), though custom or nuanced compliance rules may require human judgment, achieving near-complete automation of this workflow.
Task automatabilityclaude-sonnet-53/5AI can flag many deviations from coding/style standards and suggest fixes via static analysis and LLM code review, but full identification across complex codebases and nuanced compliance standards still requires human judgment and context setup.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or legal barriers prevent automation; compliance scanning is already standard industry practice. Minor friction exists around organizational adoption and human oversight of suggested changes, but no hard requirement for a licensed human to perform this task.
Adoption barriersclaude-sonnet-52/5No licensing requirement for this task, though organizational trust and review sign-off processes create some friction before fully replacing human QA judgment.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated compliance checking and modification suggestion has negligible marginal cost per scan (fractions of a cent), while a human QA analyst billable rate is $30–80/hour for equivalent work, representing an order-of-magnitude cost advantage for AI.
Cost vs. human wageclaude-sonnet-54/5Automated static analysis and AI review tools run at a fraction of the cost of manual review for routine deviations, though complex judgment cases still require costlier human oversight.
Technical feasibility todayclaude-haiku-4-5-202510014/5Production tools like SonarQube, Veracode, and GitHub code scanning already perform this task at scale in deployed environments, automatically flagging standard deviations and suggesting fixes. Minor gaps remain in handling novel or proprietary standards, but core functionality is mature and widely deployed.
Technical feasibility todayclaude-sonnet-53/5Linters, static analyzers, and AI code review tools (e.g., GitHub Copilot reviews, SonarQube with AI features) are deployed in production, but they have material false positive/negative rates and narrow standard coverage.

Store, retrieve, and manipulate data for analysis of system capabilities and requirements.

71

CI 55–86 · exposure 67 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Data automation is already deeply embedded in software development and QA workflows; data pipelines, automated ETL, and query automation are standard practice across tech and financial services, showing fast and mature adoption.
Sector adoption velocityclaude-sonnet-53/5Software QA and testing is a digitized, tech-forward field with growing AI tool adoption, but production-grade autonomous requirement analysis remains uncommon.
Augmentation potentialclaude-haiku-4-5-202510014/5AI systems significantly augment QA analyst productivity by automating routine data retrieval and formatting, allowing humans to focus on interpreting results and designing test strategies; tools like AI-assisted analytics and automated report generation enhance human workflow substantially.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up data querying, transformation, and pattern identification, letting analysts focus on interpreting results and requirements while staying in the loop.
Task automatabilityclaude-haiku-4-5-202510014/5Data storage, retrieval, and manipulation are largely automatable operations; modern AI systems and tools can handle SQL queries, data transformation pipelines, and database management with high consistency, achieving well over 50% time savings for structured data tasks. The main limitation is that analysis of results still often requires human judgment about what the findings mean for system capabilities.
Task automatabilityclaude-sonnet-53/5AI coding assistants can write scripts to store, query, and manipulate test data, but integrating this into specific system requirements analysis still requires human setup and validation.dec
Adoption barriersclaude-haiku-4-5-202510012/5There are minimal legal or licensing barriers to automating data operations; organizations face only modest friction from internal data governance policies or schema complexity, not regulatory or liability walls. Human oversight of data handling exists but does not require a licensed professional.
Adoption barriersclaude-sonnet-52/5No licensing requirement exists for this task, though organizational review of data integrity and system requirements creates some friction before full automation is trusted.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated database and data pipeline operations cost orders of magnitude less than hiring humans to manually store, retrieve, and transform data; cloud storage and query services are commodity-priced and far cheaper than QA analyst labor.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce time on repetitive data manipulation tasks but still require human oversight for correctness, keeping costs roughly comparable to a skilled analyst's time for non-trivial cases.
Technical feasibility todayclaude-haiku-4-5-202510015/5Deployed products (databases, ETL tools, data warehouses, and AI-assisted query/analytics systems) reliably perform these operations at scale in production across thousands of organizations daily. This is a solved, mature capability with high reliability.
Technical feasibility todayclaude-sonnet-53/5Products like GitHub Copilot, database query generators, and test data management tools exist and are used in production, but reliability varies with complex or domain-specific data schemas.

Monitor bug resolution efforts and track successes.

71

CI 61–80 · exposure 62 · augmentation 88 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams are among the fastest adopters of AI-driven monitoring and CI/CD analytics. Bug tracking automation is standard practice in agile and DevOps environments, with widespread production use across tech, finance, and large enterprises.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA teams are fast adopters of AI-driven dev-ops tooling, with widespread use of automated dashboards and bug-tracking integrations in production today.
Augmentation potentialclaude-haiku-4-5-202510014/5AI dashboards and automated alerts significantly enhance QA analysts' ability to spot trends, prioritize high-impact bugs, and make data-driven decisions. Human oversight of resolution quality remains valuable, but AI-assisted tracking raises productivity substantially on this task.
Augmentation potentialclaude-sonnet-55/5AI substantially augments this task by auto-summarizing bug trends, flagging stalled tickets, and generating status reports, letting testers focus on judgment-heavy verification.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automatically monitor bug tracking systems, aggregate resolution metrics, generate status reports, and track success rates with minimal human intervention. While some judgment about quality of fixes may require human review, the core monitoring and tracking work is highly automatable and saves well over 50% of manual effort.
Task automatabilityclaude-sonnet-53/5AI tools can automatically track bug status, correlate commits/fixes, and summarize resolution progress via integration with issue trackers, but verifying that a fix actually resolves the underlying issue and judging quality often still requires human oversight.
Adoption barriersclaude-haiku-4-5-202510012/5Minimal legal or licensing barriers exist; QA teams adopt these tools voluntarily. Some organizations prefer human validation of success metrics, but there is no regulatory requirement for a human to perform this monitoring task, and adoption friction is low.
Adoption barriersclaude-sonnet-51/5No licensing, regulatory, or human-in-the-loop legal requirement exists for this internal engineering tracking task.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven monitoring via cloud analytics or embedded tools costs a fraction of a full-time QA analyst's salary for ongoing tracking work. Integration costs are moderate and declining; the cost per monitored bug is orders of magnitude cheaper than manual daily tracking.
Cost vs. human wageclaude-sonnet-54/5Automated tracking and reporting tools are cheap to run compared to a human continuously monitoring bug trackers, though some oversight cost remains for validating fix quality.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (Jira integrations, AI-powered test analytics platforms, automated dashboards) already perform bug tracking and basic success monitoring in production. These systems reliably extract metadata, categorize issues, and flag resolution patterns, though some require configuration tuning for specific organizational contexts.
Technical feasibility todayclaude-sonnet-53/5Products like Jira automation, AI-powered dashboards, and CI/CD bots exist that track bug lifecycle and flag regressions, but comprehensive 'success tracking' with nuanced judgment is not fully reliable in production without human review.

Install, maintain, or use software testing programs.

69

CI 57–80 · exposure 62 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software and technology sectors show strong, production-level adoption of automated testing and AI-assisted testing tools; CI/CD pipelines are standard in most modern development organizations. Adoption is deep in information and fintech sectors, though some smaller or legacy organizations lag.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA functions are among the fastest-adopting domains for AI-assisted tooling, including test generation and maintenance copilots integrated into CI/CD pipelines.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assists QA professionals significantly by auto-generating test cases, identifying gaps in coverage, suggesting edge cases, and flagging anomalies in logs—all while the human QA engineer remains responsible for strategy and judgment. Productivity gains are substantial and well-documented.
Augmentation potentialclaude-sonnet-54/5AI coding assistants and test-generation tools substantially speed up writing, maintaining, and debugging test scripts, letting testers focus on strategy and edge cases while automation handles routine scaffolding.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI and tool-using agents can handle significant portions of software testing automation—test script generation, test execution, environment setup, and log analysis—achieving substantial time savings. However, some aspects like complex test design strategy and novel edge-case identification still require human judgment, preventing a full 5.
Task automatabilityclaude-sonnet-53/5Installing and maintaining commercial or open-source test frameworks (Selenium, Jest, JUnit CI pipelines) is largely scriptable and partially automatable, but ongoing configuration, environment troubleshooting, and integration with evolving codebases still require human judgment.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or legal barriers prevent automation of testing; no licensing requirement mandates human testers. Primary friction is organizational (testing culture, legacy tool integration, QA team resistance) rather than structural, making adoption relatively unobstructed.
Adoption barriersclaude-sonnet-51/5There is no licensing, regulatory, or human-contact requirement mandating a person install or maintain test software; adoption is purely a technical/organizational decision.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated testing infrastructure costs (cloud resources, tool licensing, oversight) are typically 60–80% cheaper per test execution than manual QA labor, though setup and maintenance overhead moderates the advantage. AI-driven testing further improves the ratio by reducing test maintenance burden.
Cost vs. human wageclaude-sonnet-53/5Automated testing tools reduce ongoing labor once configured, but initial setup, licensing, and required human oversight for maintenance keep costs roughly comparable to a QA engineer's time in many organizations.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature products (Selenium, Appium, AI-driven testing platforms like Testim, Katalon, Sauce Labs) reliably perform software testing installation, maintenance, and execution at scale in production environments. Minor gaps exist around handling highly custom or legacy systems, but deployment is widespread.
Technical feasibility todayclaude-sonnet-53/5Test automation tools and AI-assisted test generation/maintenance products exist and are used in production, but they still require engineers to set up, debug flaky tests, and interpret failures rather than running fully autonomously.

Identify, analyze, and document problems with program function, output, online screen, or content.

68

CI 55–81 · exposure 62 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Tech-sector QA has rapidly and deeply adopted automation for decades; continuous integration/deployment pipelines rely heavily on automated testing. Production adoption is mainstream rather than experimental in software development organizations.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA teams in tech-forward companies are adopting AI-assisted testing tools moderately fast, but many organizations still rely primarily on manual or script-based testing with pilots rather than full production reliance.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assists QA analysts by automating repetitive test execution and documenting findings, freeing humans to focus on exploratory testing, edge cases, and quality strategy. Tools like test-generation and intelligent test prioritization meaningfully raise human productivity while keeping the analyst in the loop.
Augmentation potentialclaude-sonnet-54/5AI tools substantially speed up defect detection, log analysis, and report drafting, letting testers focus on judgment-heavy edge cases while the human remains responsible for final validation.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automate significant portions of QA work through automated testing frameworks, visual regression detection, and log analysis, achieving >50% time savings on routine test execution and basic defect identification. However, complex edge cases and novel functionality still require human judgment, preventing a full 5-rating.
Task automatabilityclaude-sonnet-53/5AI can run tests, flag anomalies, and draft bug reports, but reliably identifying subtle functional defects, root-causing them, and documenting them with proper context still requires human judgment for many real-world cases.A large portion of the mechanical detection/documentation work is automatable, but not the full end-to-end task at equal quality.
Adoption barriersclaude-haiku-4-5-202510012/5QA automation faces minimal legal or regulatory barriers and does not require human sign-off by law. Organizational friction exists (legacy test suite maintenance, false positives requiring human triage) but these are operational friction rather than hard blockers.
Adoption barriersclaude-sonnet-52/5No licensing or legal sign-off is required for QA analysis, though some regulated industries (e.g., medical device software) impose documentation and validation requirements that add friction.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated testing infrastructure costs a fraction of manual QA labor per test execution, and once configured, scales to thousands of test runs at negligible marginal cost compared to hiring QA testers, achieving orders-of-magnitude cost advantage.
Cost vs. human wageclaude-sonnet-53/5AI-assisted testing tools reduce manual effort significantly, but integration, maintenance, and human oversight of AI-flagged issues keep total costs roughly comparable to skilled QA labor rather than an order of magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature products (Selenium, Playwright, visual regression tools, and AI-powered test generation platforms like Testim and Applitools) reliably perform automated testing in production at scale. Deployed AI systems can document issues, though some human validation remains common practice.
Technical feasibility todayclaude-sonnet-53/5Deployed tools (AI-assisted test automation, log analysis, visual regression detection, LLM-based bug triage) exist and are used in production, but they still have notable false-positive/negative rates and require human review for complex or novel issues.

Update automated test scripts to ensure currency.

68

CI 61–75 · exposure 58 · augmentation 88 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams have rapidly adopted AI coding assistants; use of AI for test automation and script updates is now commonplace in information-sector companies and startups.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are high-digitization, fast-adopting sectors where AI coding tools have seen rapid production deployment over the last two years.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered code assistants demonstrably accelerate script writing and maintenance by suggesting fixes, refactoring, and deprecation updates. The human QA engineer reviews and validates, creating strong productivity gains while preserving control.
Augmentation potentialclaude-sonnet-55/5AI coding assistants substantially speed up the process of locating outdated scripts, suggesting fixes, and regenerating assertions, while a human tester still validates correctness.
Task automatabilityclaude-haiku-4-5-202510013/5Updating test scripts requires understanding intent, code modification, and validation. AI can handle routine script repairs and syntax updates (~50% of work), but context-dependent changes and quality assurance still need human judgment.
Task automatabilityclaude-sonnet-54/5Updating existing test scripts (e.g., adjusting selectors, API calls, or assertions to match code changes) is a pattern-matching/code-editing task that LLM-based coding agents handle well, especially with access to diffs and test failure logs.itle for at least half the effort with equal or better quality.
Adoption barriersclaude-haiku-4-5-202510012/5No licensing requirement to automate script updates. Organizational friction exists (teams prefer human review of test logic), but technical or legal barriers to deployment are minimal.
Adoption barriersclaude-sonnet-51/5There is no licensing or regulatory requirement that a human must update test scripts, and organizations readily use automated tools for this maintenance work.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference for code generation is cheap, and integration costs are low in typical dev environments. The per-update cost is materially lower than paying a QA analyst for routine script maintenance, though oversight adds overhead.
Cost vs. human wageclaude-sonnet-54/5Inference cost for script updates is minor compared to a QA engineer's hourly cost, though integration and review overhead reduce the savings somewhat, making the gap large but not the full order-of-magnitude in all cases.
Technical feasibility todayclaude-haiku-4-5-202510013/5Tools like GitHub Copilot and specialized code-generation AI perform script repairs and updates in production, but with material error rates and gaps in complex scenarios. Deployed products exist but require significant human oversight.
Technical feasibility todayclaude-sonnet-53/5AI coding assistants (Copilot, Cursor, Codeium) and test-maintenance tools are deployed in production and can auto-update scripts, but they still require human review for logic changes and often miss context-specific edge cases, so error rates remain non-trivial.

Evaluate or recommend software for testing or bug tracking.

67

CI 55–79 · exposure 62 · augmentation 88 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software and IT organizations are rapidly adopting AI for procurement and tool evaluation decisions; major tech firms and professional services companies routinely use generative AI to assess tool options, with visible adoption in production consulting and internal IT processes.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA teams are moderately fast adopters of AI tools for research and documentation tasks, but tool selection processes remain largely human-driven with AI used as a research aid rather than a replacement.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments QA analysts by generating feature comparisons, cost analyses, and integration assessments in minutes rather than hours of manual research, while analysts retain judgment about organizational priorities, risk tolerance, and team preferences.
Augmentation potentialclaude-sonnet-54/5AI significantly speeds up gathering comparative information, summarizing feature sets, and drafting recommendation documents, meaningfully boosting analyst productivity even though final judgment remains human.
Task automatabilityclaude-haiku-4-5-202510014/5AI can substantially automate the evaluation and comparison of testing and bug tracking tools by analyzing features, pricing, integrations, and use-case fit from documentation and vendor materials, achieving significant time savings. However, final selection typically requires organizational judgment about workflow fit and legacy system compatibility, preventing full end-to-end automation.
Task automatabilityclaude-sonnet-53/5AI can research, compare, and summarize software testing/bug-tracking tools and even draft recommendation reports, but final vendor selection involves organizational context, budget negotiation, and stakeholder judgment that still requires human decision-making.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or legal barriers exist to automating tool evaluation and recommendation; organizational purchasing processes typically involve multiple stakeholders, but AI outputs serve as informational input rather than binding decisions, creating only modest friction.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates a human evaluator, though internal procurement policies and vendor relationship management create some organizational friction.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference cost for evaluating and comparing tools is minimal (cents to low dollars per evaluation), while a QA analyst's time for thorough tool assessment, research, and recommendation costs hundreds of dollars, making AI at least an order of magnitude cheaper all-in.
Cost vs. human wageclaude-sonnet-53/5AI can cheaply produce comparative research and draft recommendations, but human validation, vendor demos, and internal buy-in still require meaningful analyst time, keeping overall cost savings moderate rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature products like Claude, ChatGPT, and specialized code-analysis tools reliably evaluate software tools, generate comparison matrices, and identify gaps in tool feature sets against stated requirements. Deployed AI systems in enterprises already assist with tool evaluation in production environments with acceptable reliability for recommendation support.
Technical feasibility todayclaude-sonnet-53/5Products like ChatGPT, Copilot, and specialized research assistants can generate credible tool comparisons today, but no deployed system reliably owns the full evaluation-to-recommendation workflow within an organization's procurement process.

Test system modifications to prepare for implementation.

66

CI 57–75 · exposure 62 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and DevOps-heavy sectors show rapid, deep adoption of AI-driven testing tools and continuous integration pipelines. Test automation is now standard practice in tech, finance, and digital-first enterprises, with high production deployment rates.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are among the fastest-adopting sectors for AI tooling, with CI/CD pipelines increasingly incorporating AI-assisted test generation and anomaly detection in production.
Augmentation potentialclaude-haiku-4-5-202510014/5AI testing tools significantly enhance QA analyst productivity by automating routine test execution, generating test cases, and flagging anomalies, while humans focus on test design, edge-case discovery, and complex validation logic. The human-AI partnership is highly productive and widely practiced.
Augmentation potentialclaude-sonnet-54/5AI significantly speeds up test case generation, script writing, and defect triage, letting testers focus on exploratory testing and judgment calls while staying in the loop.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI can automate significant portions of system testing through automated test generation, execution, and regression testing, achieving substantial time savings. However, complex integration testing and validation of business logic modifications often requires human judgment, preventing full end-to-end automation at the 50% threshold consistently.
Task automatabilityclaude-sonnet-53/5AI can generate and execute test cases, run regression suites, and flag anomalies for many modifications, but complex system-level integration testing still requires human judgment on risk areas, edge cases, and business context.//This gets roughly half automated with substantial setup rather than full end-to-end coverage.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or licensing barriers exist for automated testing; organizations maintain internal control and can retain human review where needed. QA automation is widely adopted without legal mandates requiring human sign-off on test execution itself.
Adoption barriersclaude-sonnet-52/5No licensing requirement exists for this task, though some regulated industries (healthcare, finance, aviation software) impose validation and audit requirements that create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated testing infrastructure is significantly cheaper than manual QA labor per test cycle executed, especially for regression suites. Upfront setup and maintenance costs are offset quickly by the volume of tests run, making AI testing roughly 5-10x cheaper than equivalent human testing effort.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce test-writing time but require infrastructure, maintenance, and human review to avoid false positives/negatives, keeping costs roughly comparable to skilled QA labor when done properly.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature testing automation products (Selenium, TestComplete, AI-enhanced tools like Applitools, Mabl) reliably perform regression testing and UI testing in production at scale. End-to-end test orchestration is increasingly deployed, though some edge cases and novel system architectures still require human oversight.
Technical feasibility todayclaude-sonnet-53/5AI-assisted test generation, test automation frameworks, and coverage analysis tools are deployed in production at many software firms, but reliability varies by codebase complexity and often requires human curation of test suites.

Perform initial debugging procedures by reviewing configuration files, logs, or code pieces to determine breakdown source.

66

CI 57–75 · exposure 62 · augmentation 100 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams, predominantly in information/tech sectors, have rapidly adopted AI-assisted debugging tools; plugins, integrated development environments, and CI/CD pipelines increasingly incorporate AI analysis by default.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are among the fastest-adopting fields for AI coding/debugging assistants, with widespread pilot and production use of tools like Copilot, Cursor, and AI-augmented observability platforms.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments human debuggers by surfacing candidates, narrowing search space, and flagging anomalies in logs and configs in real time, allowing QA analysts to focus on interpretation and complex root-cause analysis rather than manual review.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up initial triage by summarizing logs, flagging anomalies, and suggesting likely faulty code sections, letting testers focus on verification and complex judgment calls.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems (static analyzers, LLM-based code reviewers, log analysis tools) can automate a substantial portion of initial debugging by identifying common issues in configuration files, logs, and code patterns with high accuracy, achieving >50% time savings on routine diagnostics. However, complex or context-dependent breakdowns often require human judgment, preventing a full 5 rating.
Task automatabilityclaude-sonnet-53/5AI coding assistants can analyze logs, config files, and code to suggest likely breakdown sources, but complex or novel failures still require human judgment and system context, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510012/5Few legal or regulatory barriers exist; QA automation is standard practice and does not require human licensing. Some organizational friction and preference for human review in critical systems apply, but substitution is technically and contractually unconstrained.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates a human debugger, though organizational trust in AI-driven root-cause analysis for production systems creates some friction, especially for critical systems.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-based debugging and log-analysis services cost substantially less per incident than a senior QA analyst's time, especially for routine or repetitive failure patterns. Inference and integration costs are low relative to human salary and time investment.
Cost vs. human wageclaude-sonnet-53/5AI tools can cheaply scan logs and code at scale, but the need for human oversight, iterative investigation, and integration with existing debugging workflows keeps costs roughly comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature products like GitHub Copilot, SonarQube, Splunk, and specialized log-analysis AI are deployed in production across many organizations and reliably identify configuration errors and common code issues. Minor gaps remain in novel failure modes, but the core task is well-covered by existing tools.
Technical feasibility todayclaude-sonnet-53/5Products like GitHub Copilot, Sentry AI, and various log-analysis/observability tools with LLM integration perform this in production, but they often surface candidate causes rather than definitively diagnosing issues, requiring engineer verification.

Design test plans, scenarios, scripts, or procedures.

58

CI 55–61 · exposure 50 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams operate in highly digitized, fast-moving sectors with strong early adoption of AI coding assistants; many organizations are already integrating AI into test automation pipelines.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA are within tech/professional services, a fast-adopting sector, but AI-driven test design is still often pilot-stage rather than the default replacing human-authored test plans.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly augments QA engineers by rapidly generating test scenarios, identifying edge cases, and drafting test scripts, allowing analysts to focus on risk prioritization and validation rather than manual scripting.
Augmentation potentialclaude-sonnet-54/5AI substantially accelerates drafting of test scripts, generating edge cases, and suggesting scenarios based on code or specs, meaningfully boosting QA analyst productivity while they retain oversight and judgment.
Task automatabilityclaude-haiku-4-5-202510013/5AI can generate substantial portions of test plans, scenarios, and scripts (especially boilerplate and common cases) with significant time savings, but designing comprehensive test strategies often requires domain expertise, risk assessment, and understanding of business requirements that demand human judgment to ensure adequate coverage.
Task automatabilityclaude-sonnet-53/5AI can draft test plans and generate test cases from requirements or code, but designing comprehensive scenarios that reflect real user behavior, edge cases, and business risk still requires significant human judgment and validation.
Adoption barriersclaude-haiku-4-5-202510012/5Few regulatory or legal barriers prevent AI-assisted test design; companies routinely adopt AI coding tools, and test plan generation does not require licensed practitioners or final sign-off by law.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates human authorship of test plans, though organizational risk tolerance and quality assurance culture create some friction against fully automated design in critical systems.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-assisted test design is significantly cheaper than hiring QA analysts to write all plans and scripts from scratch; inference costs are low relative to senior QA labor costs, though integration and validation oversight still require human time.
Cost vs. human wageclaude-sonnet-53/5AI-assisted test generation reduces time on repetitive scripting, but the need for human review, refinement, and domain-specific scenario design keeps overall costs roughly comparable to a skilled QA analyst working with AI tools.
Technical feasibility todayclaude-haiku-4-5-202510013/5Products like GitHub Copilot, ChatGPT, and specialized test generation tools can produce test scripts and procedures reliably, but they often generate incomplete or redundant test cases and require human review to validate coverage and alignment with actual product risks.
Technical feasibility todayclaude-sonnet-53/5Products like GitHub Copilot, test-generation tools, and AI-powered QA platforms (e.g., Testim, Functionize) exist and are used in production, but they typically handle scaffolding or partial test generation rather than full end-to-end test plan design reliably.

Develop testing programs that address areas such as database impacts, software scenarios, regression testing, negative testing, error or bug retests, or usability.

57

CI 57–57 · exposure 50 · augmentation 100 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams are rapidly adopting AI-assisted testing tools in production settings, with test generation and automation frameworks becoming standard in tech-forward organizations. Adoption is particularly fast in information and software sectors where ROI is clear and integration friction is low.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are high-digitization, fast-adopting sectors where AI-assisted testing tools are increasingly integrated into CI/CD pipelines in production environments.
Augmentation potentialclaude-haiku-4-5-202510015/5AI excels at augmenting QA analysts by generating test case suggestions, identifying edge cases, automating regression test discovery, and drafting test plans that humans refine. This significantly raises the productivity of human testers while keeping them in control of strategy and final test suite design.
Augmentation potentialclaude-sonnet-55/5AI significantly boosts productivity by auto-generating test cases, identifying edge cases, and suggesting regression scenarios, while human testers retain oversight for judgment-heavy areas like usability and negative testing design.
Task automatabilityclaude-haiku-4-5-202510013/5AI can generate test cases, outline testing scenarios, and draft testing programs with reasonable quality, but designing comprehensive testing strategies that balance coverage, risk, and resource constraints typically requires human judgment about domain-specific risks and business priorities. Roughly half the work of creating a testing program could be automated with current tools.
Task automatabilityclaude-sonnet-53/5AI coding assistants can draft test plans, generate test cases, and suggest coverage for regression/negative scenarios, but designing a comprehensive testing program requires domain judgment, risk prioritization, and system-specific knowledge that still needs substantial human oversight.dfg
Adoption barriersclaude-haiku-4-5-202510012/5Testing programs are typically owned by organizations without strict licensing or regulatory barriers on automation itself, though team practices and internal governance may prefer human review. Adoption is primarily constrained by organizational friction and confidence in AI-generated test quality rather than hard legal requirements.
Adoption barriersclaude-sonnet-52/5No licensing requirement for QA testing, but organizational risk tolerance and the need for domain-specific validation create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-powered test generation can reduce manual effort significantly, making it cost-competitive with human testers on routine test case creation, but oversight, validation, and scenario design still require skilled QA labor. Overall cost is roughly comparable when accounting for integration and quality assurance.
Cost vs. human wageclaude-sonnet-53/5AI-assisted test generation reduces some manual effort and cost, but human review, environment setup, and integration into CI/CD pipelines keep overall costs roughly comparable to human-led efforts for full program design.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI-assisted test generation and planning tools exist in production (e.g., code-aware AI that suggests test cases), but fully autonomous development of enterprise-grade testing programs with high reliability remains limited. Products perform parts of this task well but not end-to-end with consistent quality across diverse scenarios.
Technical feasibility todayclaude-sonnet-53/5Deployed tools (Copilot, test-generation platforms) assist in creating unit/integration tests and some regression suites in production, but comprehensive test program design across database, usability, and negative testing is not yet reliably automated end-to-end.

Design or develop automated testing tools.

57

CI 57–57 · exposure 50 · augmentation 88 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Tech and software development sectors are rapidly adopting AI-assisted code generation and test automation; major companies deploy Copilot and similar tools at scale in QA workflows. Adoption is faster than average but not yet universal.
Sector adoption velocityclaude-sonnet-54/5Software engineering and QA are among the fastest AI-adopting domains, with AI-assisted coding and test generation tools rapidly integrated into CI/CD pipelines industry-wide.
Augmentation potentialclaude-haiku-4-5-202510014/5AI code completion and test case generation significantly boost QA engineer productivity by automating boilerplate and suggesting test scenarios, allowing engineers to focus on edge cases, architecture, and validation logic while staying in the loop.
Augmentation potentialclaude-sonnet-55/5AI substantially accelerates writing test scripts, generating edge cases, and scaffolding automation frameworks, making it a strong productivity multiplier for QA engineers who remain responsible for design and validation.
Task automatabilityclaude-haiku-4-5-202510013/5AI can assist with code generation for test scripts and test case design, but full end-to-end development of sophisticated testing frameworks requires architectural decisions, integration with legacy systems, and domain-specific logic that currently demand human oversight. Partial automation with significant human involvement is achievable.
Task automatabilityclaude-sonnet-53/5AI coding assistants can generate test scripts, frameworks, and automation harnesses from specifications, but designing robust, maintainable test automation architecture for complex systems still requires significant human judgment and iteration.
Adoption barriersclaude-haiku-4-5-202510012/5Testing tool development is a standard software engineering task with no regulatory or legal requirement for human certification. Organizational friction around tool validation and trust exists but is modest compared to regulated domains.
Adoption barriersclaude-sonnet-52/5No licensing or regulatory requirement mandates a human perform this task, though organizational quality standards and code review practices create some friction to full automation.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-assisted code generation reduces development time for routine test scripts, but setup, prompt engineering, validation, and integration overhead make total costs roughly comparable to experienced QA engineers writing tests manually on many projects.
Cost vs. human wageclaude-sonnet-53/5AI reduces time spent writing boilerplate test code and scripts, but integration, validation, and maintenance of automation frameworks still require substantial engineering oversight, keeping costs roughly comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510013/5LLM-based code generation tools (Copilot, Claude) and dedicated test generation products exist and are in production use, but they require substantial human review, refinement, and debugging. Error rates and scope limitations mean they function reliably only within narrow, well-defined test scenarios.
Technical feasibility todayclaude-sonnet-53/5Deployed tools like GitHub Copilot, Testim, and AI-augmented test generation platforms are used in production, but they typically handle scaffolding and partial test-case generation rather than full end-to-end tool design with reliability at scale.

Install and configure recreations of software production environments to allow testing of software performance.

57

CI 38–77 · exposure 50 · augmentation 75 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Software and tech sectors have rapidly and deeply adopted Infrastructure-as-Code, automated deployment, and CI/CD practices over the past decade. This is a high-adoption area with widespread production use in DevOps and QA teams.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA teams are moderately fast adopters of AI-assisted DevOps tooling, though full environment replication automation remains a partial practice.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-driven environment setup and configuration tools substantially augment QA engineers by handling boilerplate provisioning, freeing them to focus on complex environment validation, performance tuning, and troubleshooting—a strong productivity multiplier.
Augmentation potentialclaude-sonnet-54/5AI coding assistants and infrastructure automation tools significantly speed up writing configuration scripts, generating IaC templates, and troubleshooting setup issues, boosting engineer productivity.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automate most of this task—automated deployment tools, Infrastructure-as-Code generation, and containerization platforms can recreate production environments from configuration files with minimal human intervention, achieving well over 50% time savings. However, some environment-specific customization and validation may still require human oversight.
Task automatabilityclaude-sonnet-52/5Environment setup involves infrastructure-as-code, dependency management, and troubleshooting idiosyncratic production configs that current AI can assist with but not reliably execute end-to-end without human oversight.'
Adoption barriersclaude-haiku-4-5-202510012/5Few legal or regulatory barriers exist; organizations may prefer human oversight for critical systems, but nothing legally mandates human configuration. Adoption is hindered mainly by organizational inertia and legacy system constraints rather than structural barriers.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but organizational risk of misconfigured test environments causing downstream defects creates moderate caution and review requirements.
Cost vs. human wageclaude-haiku-4-5-202510014/5Infrastructure automation tools are inexpensive relative to the loaded wage of a QA engineer manually provisioning and configuring environments, particularly for repetitive multi-environment setups. Cost scales favorably with reuse and standardization.
Cost vs. human wageclaude-sonnet-52/5AI can cut scripting time but debugging environment-specific issues (networking, licensing, legacy systems) still requires skilled engineer time, keeping costs comparable to human effort in most cases.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature products demonstrably perform this in production: Terraform, Ansible, Kubernetes, Docker, and CI/CD platforms reliably automate environment setup and configuration at scale across many organizations. Error rates are low for standard infrastructure patterns.
Technical feasibility todayclaude-sonnet-52/5Tools like Docker/Terraform automation and AI coding assistants can generate config scripts, but no deployed product reliably stands up full production-mirroring test environments autonomously across diverse stacks.

Provide feedback and recommendations to developers on software usability and functionality.

54

CI 51–57 · exposure 50 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development and QA teams, concentrated in high-digitization sectors (tech, finance, SaaS), have rapidly adopted automated testing tools and AI-assisted analysis in production. Tools like ChatGPT for code review and test generation are now mainstream in many organizations, signaling fast, measurable adoption.
Sector adoption velocityclaude-sonnet-53/5Software development and QA are relatively tech-forward, with growing use of AI-assisted code review and testing tools, but full delegation of usability feedback to AI remains uncommon in production workflows.
Augmentation potentialclaude-haiku-4-5-202510014/5AI excels at augmenting QA analysts: generating initial test cases, surfacing patterns in user feedback, cross-referencing code changes with potential impact zones, and drafting recommendations that analysts then refine and validate. This high-leverage assistance substantially raises analyst productivity while maintaining human oversight.
Augmentation potentialclaude-sonnet-54/5AI significantly boosts QA analysts' productivity by flagging usability issues, summarizing bug patterns, and drafting recommendations, letting humans focus on judgment calls and communication with developers.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate portions of this task (automated testing, static analysis, and user feedback pattern analysis), but providing nuanced recommendations on usability and functionality improvements typically requires human judgment, context understanding, and domain knowledge. A hybrid workflow achieves roughly 40–50% efficiency gain, not decisively crossing the equal-quality ≥50% threshold.
Task automatabilityclaude-sonnet-53/5AI can analyze test results, logs, and usability data to draft feedback and suggest improvements, but synthesizing nuanced UX judgment and prioritizing recommendations in context still needs human review for most of the task's value.
Adoption barriersclaude-haiku-4-5-202510012/5Few hard regulatory or licensing barriers prevent automation of feedback generation, but organizational inertia and developer skepticism of AI-generated recommendations without human validation create meaningful friction. Some teams require explicit human sign-off on critical usability findings.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates a human provide this feedback, though organizational trust and quality assurance norms create some friction against fully automated recommendations.
Cost vs. human wageclaude-haiku-4-5-202510012/5Automated testing and AI-assisted feedback tools are considerably cheaper per test case than hiring QA analysts, but the overhead of prompt engineering, output validation, and human review to refine AI-generated recommendations narrows the cost advantage. All-in AI cost remains roughly comparable to or slightly cheaper than a mid-level QA salary, not an order of magnitude better.
Cost vs. human wageclaude-sonnet-53/5AI tools can cheaply generate draft feedback, but integration, verification, and follow-up communication with developers still require analyst time, keeping the overall cost roughly comparable to a human doing the full task well.
Technical feasibility todayclaude-haiku-4-5-202510013/5Mature products (Testim, Applitools, BrowserStack) and large language models can generate usability feedback and code analysis recommendations at production scale, but they frequently miss domain-specific issues, user intent subtleties, and cross-feature interactions that human testers catch. Error rates and recommendation quality gaps persist.
Technical feasibility todayclaude-sonnet-53/5Products like AI code review assistants and QA copilots (e.g., Copilot, specialized QA tools) generate usability/functionality feedback today, but they operate with narrow scope and require human validation before being actionable.

Review software documentation to ensure technical accuracy, compliance, or completeness, or to mitigate risks.

54

CI 50–59 · exposure 50 · augmentation 88 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Tech and financial services sectors are piloting AI-assisted documentation review, but widespread production adoption remains limited; most organizations still rely primarily on manual review with emerging tool experimentation. Adoption is accelerating in digitally mature sectors but lags in traditional industries.
Sector adoption velocityclaude-sonnet-53/5Software/tech sectors adopt AI tools quickly, but documentation review specifically is still in a pilot/emerging tool phase rather than deeply embedded standard practice across QA teams.
Augmentation potentialclaude-haiku-4-5-202510014/5AI powerfully augments QA analysts by automating tedious checks (missing fields, formatting, cross-reference validation) and flagging anomalies, freeing human effort for nuanced technical accuracy and risk judgment. Tools demonstrably increase review speed and consistency when humans remain responsible for final assessment.
Augmentation potentialclaude-sonnet-55/5AI is highly effective at flagging inconsistencies, missing sections, ambiguous language, and cross-referencing documentation against code or specs, substantially speeding up human reviewers while they retain final judgment.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate parts of documentation review—checking for missing sections, consistency issues, and basic compliance patterns—but requires human judgment for technical accuracy validation against complex systems and risk assessment nuance. Achieving 50% time savings is feasible with current tools, though full end-to-end automation faces limitations in contextual understanding.
Task automatabilityclaude-sonnet-53/5LLMs can review documentation for technical accuracy, consistency, and completeness against specs, but nuanced risk mitigation and domain judgment still require human verification, so only partial time savings are realized end-to-end.
Adoption barriersclaude-haiku-4-5-202510013/5Documentation review sits in a gray zone: regulatory frameworks (SOC 2, ISO, healthcare compliance) often require demonstrated due diligence, and organizations may prefer human sign-off for liability reasons, but no strict licensing prevents AI-assisted or AI-led review. Customer expectations and risk-aversion create friction without legal prohibition.
Adoption barriersclaude-sonnet-52/5No licensing requirement mandates a human review documentation, though regulated industries (medical devices, aerospace, finance) may require sign-off by a qualified engineer or QA lead for compliance purposes.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI inference cost for documentation analysis is low, but integration overhead and required human oversight for quality assurance add meaningful expense. The all-in cost approaches parity with a mid-level QA analyst reviewing the same volume, with advantages emerging only at scale.
Cost vs. human wageclaude-sonnet-54/5AI-assisted review of large documentation sets is far cheaper per page than manual analyst review, though some human oversight is still needed to catch domain-specific compliance issues, keeping it just short of a full order-of-magnitude saving in practice.
Technical feasibility todayclaude-haiku-4-5-202510013/5Products like AI-powered documentation analysis tools and LLM-based compliance checkers exist and are deployed in some organizations, but they produce false positives/negatives at material rates and typically require human review for validation. Production use is growing but not yet mature at scale across the industry.
Technical feasibility todayclaude-sonnet-53/5Products like AI-assisted doc linters, code-doc consistency checkers, and LLM-based reviewers exist and are used in some QA pipelines, but they are not yet standard, reliable production tools for comprehensive documentation review across compliance and risk dimensions.

Modify existing software to correct errors, allow it to adapt to new hardware, or to improve its performance.

53

CI 49–57 · exposure 50 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Software development is a highly digitized, information-intensive sector with rapid AI adoption; GitHub Copilot, ChatGPT-assisted coding, and AI-powered testing frameworks are already in widespread use in many technology organizations. Pilots and production adoption are common, particularly in larger tech-forward firms.
Sector adoption velocityclaude-sonnet-54/5Software engineering is among the fastest-adopting sectors for AI coding tools, with widespread production use of AI-assisted code editing and bug fixing at scale.
Augmentation potentialclaude-haiku-4-5-202510014/5AI code generation, bug detection suggestions, and test case generation substantially assist developers in code modification workflows, raising their productivity without removing them from the loop. Developers remain responsible for validation and architecture, but AI-suggested fixes and refactorings demonstrably accelerate iteration.
Augmentation potentialclaude-sonnet-55/5AI substantially accelerates locating bugs, suggesting fixes, and drafting performance optimizations, while engineers retain responsibility for validation and final decisions.
Task automatabilityclaude-haiku-4-5-202510013/5AI can assist with identifying error patterns, suggesting code fixes, and generating patches, but modifying software to adapt to new hardware or optimize performance typically requires domain knowledge about system architecture and integration testing that current tools handle inconsistently. Roughly half the task—bug fix suggestions and performance improvement proposals—can be automated with meaningful time savings, while the other half (validation, integration, hardware-specific adaptation) often requires human expertise.
Task automatabilityclaude-sonnet-53/5AI coding assistants can generate bug fixes and refactors for well-scoped issues, but complex debugging across large codebases, hardware adaptation, and performance tuning still require significant human judgment and verification.
Adoption barriersclaude-haiku-4-5-202510013/5Software modifications affecting production systems typically require human developer review and sign-off for liability and quality assurance reasons, creating organizational friction. Regulatory requirements (e.g., in safety-critical or financial systems) may mandate human validation, though general business software faces fewer formal barriers.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but organizational risk aversion around production code changes, code review gates, and liability for introduced bugs create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510013/5Inference costs for LLM-based code generation and AI-assisted testing are dropping, but integration, validation infrastructure, and human oversight overhead offset much of the savings. The all-in cost is becoming comparable to junior developer time but remains above that of fully automated commodity tasks.
Cost vs. human wageclaude-sonnet-52/5AI tools reduce time on simple fixes but complex modifications still require senior engineer oversight, code review, and testing infrastructure, keeping all-in costs closer to human-comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510013/5Code completion and bug-detection tools like GitHub Copilot and static analyzers are deployed at scale, but they still produce errors, require substantial human review, and struggle with complex architectural changes or hardware-specific modifications. Products exist and perform parts of this task reliably in production, but material error rates and the need for human sign-off prevent full reliability.
Technical feasibility todayclaude-sonnet-53/5Products like Copilot, Cursor, and AI-assisted debugging tools are deployed in production and help with code fixes, but reliability drops sharply for non-trivial bugs, performance regressions, or hardware-specific issues.

Investigate customer problems referred by technical support.

42

CI 32–51 · exposure 38 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Tech and software companies are piloting AI-assisted QA analysis and log tools, but widespread production deployment of autonomous investigation at scale remains limited. Adoption is in the middling stage: common in forward-facing teams, but not yet standard practice.
Sector adoption velocityclaude-sonnet-53/5Software/tech sector has high general AI adoption, with AI-assisted debugging and support ticket triage tools increasingly used, though full investigative work remains largely human-led.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly assist QA analysts by auto-summarizing logs, suggesting likely root causes, and flagging anomalies, allowing analysts to focus on validation and complex troubleshooting. This augmentation materially raises productivity while the analyst retains decision authority.
Augmentation potentialclaude-sonnet-54/5AI tools can significantly speed up log analysis, error pattern recognition, and generating hypotheses for root cause, greatly aiding the analyst's investigation process.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate parts of investigation—log analysis, pattern detection, and initial categorization—but resolving customer problems typically requires domain expertise, judgment about context, and understanding of non-technical factors that affect reproduction and remediation. Roughly half the investigative workflow could be automated with setup.
Task automatabilityclaude-sonnet-52/5Investigating customer-reported bugs requires reproducing issues, reading logs, understanding system context, and judgment about root cause, which AI can assist but not reliably complete end-to-end today.rgo
Adoption barriersclaude-haiku-4-5-202510013/5Customers often require a human touch and accountability; organizations need oversight of AI-driven investigation to ensure quality and liability. Technical skill and organizational trust in automation create moderate friction, though no hard legal barrier prevents AI assistance.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but organizational trust, need for domain knowledge of proprietary systems, and accountability for correct root-cause diagnosis create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5Investigation tools are cheaper per task component, but total cost including integration, oversight for accuracy, and the human review required to validate findings keeps the ratio closer to parity than a decisive cost advantage.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply triage and summarize logs, but full investigation still requires human QA engineers to validate and reproduce issues, keeping overall cost comparable to human labor.
Technical feasibility todayclaude-haiku-4-5-202510013/5Products exist for automated log analysis, anomaly detection, and defect categorization, but they operate with material error rates and require significant human review. No mature end-to-end system reliably investigates and resolves customer problems independently at production scale.
Technical feasibility todayclaude-sonnet-52/5Some AI-assisted debugging and log-analysis tools exist, but no deployed product reliably investigates and resolves ambiguous customer-reported issues autonomously in production.

Coordinate user or third-party testing.

35

CI 32–38 · exposure 25 · augmentation 63 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Software development has moderate AI adoption, with pilots of test automation tools common; however, human coordination of distributed testing teams remains the norm, not an area seeing rapid displacement.
Sector adoption velocityclaude-sonnet-53/5Software/QA teams are moderately fast adopters of AI for test management and communication tools, though coordination-specific automation remains a pilot-level use case.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist by auto-scheduling, tracking test status, and flagging delays, allowing human coordinators to focus on priority conflicts and stakeholder communication—meaningful but not transformative support.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by drafting communications, tracking test schedules, summarizing tester feedback, and flagging issues, improving efficiency while a human still manages relationships.
Task automatabilityclaude-haiku-4-5-202510012/5AI can help schedule tests and aggregate results, but coordinating human testers requires negotiation, context-switching, and relationship management that current systems cannot reliably handle end-to-end. Significant manual oversight remains necessary.
Task automatabilityclaude-sonnet-52/5Coordinating people—scheduling, communicating with users/third parties, tracking feedback—requires interpersonal negotiation and judgment that current AI cannot fully replace, though parts like scheduling or summarizing feedback can be assisted.
Adoption barriersclaude-haiku-4-5-202510013/5No strict legal barrier prevents some automation of coordination logistics, but organizational practice strongly favors a human point-person for accountability and conflict resolution among testing partners.
Adoption barriersclaude-sonnet-52/5No licensing requirement, but organizational trust, client relationship management, and accountability for coordinating external parties create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Coordination tools cost substantial per-user licensing and integration overhead, and the human coordinator's judgment and relationship-building remain irreplaceable, making AI cheaper only for narrow task slices like scheduling.
Cost vs. human wageclaude-sonnet-52/5Human coordinators are still needed to interface with external stakeholders and resolve ambiguities, so AI tools reduce some overhead but don't yet replace the labor cost outright.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can send automated reminders and organize test data, no deployed product reliably coordinates the full workflow of multiple human testers across different schedules and priorities. Existing tools provide partial support but require constant human mediation.
Technical feasibility todayclaude-sonnet-52/5No deployed product autonomously manages third-party test coordination end-to-end; existing tools (project trackers, chatbots) only handle fragments like reminders or status updates.

Plan test schedules or strategies in accordance with project scope or delivery dates.

34

CI 30–38 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Most QA teams still rely on manual or spreadsheet-based scheduling with limited AI integration in production. Adoption of dedicated AI planning tools remains in pilot/early stages; most organizations have not shifted planning tasks to automated systems.
Sector adoption velocityclaude-sonnet-53/5Software development and QA sectors show moderate AI tool adoption for planning and test generation, though full strategic test scheduling automation is still mostly pilot-stage.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by analyzing project data, suggesting test timeline templates, flagging scope conflicts, and automating administrative scheduling tasks, allowing QA leads to focus on strategic prioritization and stakeholder alignment.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by suggesting test coverage priorities, timelines, and risk areas based on project data, significantly speeding up the planning process while humans finalize decisions.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in analyzing project data and suggesting test timelines, creating effective test schedules requires understanding complex project dynamics, resource constraints, and risk prioritization that currently demand substantial human judgment. End-to-end automation with 50% time savings at equal quality is not demonstrated by current systems.
Task automatabilityclaude-sonnet-52/5Planning test schedules requires integrating project scope, resource constraints, delivery deadlines, and risk prioritization judgments that current AI can assist but not fully own end-to-end."},"feasibility":{"rating":2,"rationale":"Some project-management and test-planning tools offer AI-assisted scheduling suggestions, but no mature product reliably generates full test strategies aligned to scope and deadlines without heavy human curation."}, "placeholder":true
Adoption barriersclaude-haiku-4-5-202510013/5Test planning is typically embedded in organizational processes and project governance where human QA leads are expected to own accountability for schedules and scope alignment. Some friction from process integration exists, though no formal licensing requirement blocks automation.
Adoption barriersclaude-sonnet-52/5No licensing or regulatory requirement mandates human sign-off, but organizational reliance on human judgment for scheduling trade-offs creates moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI planning assistants require significant setup, domain expertise to configure, and human validation of outputs. The cost of integration, ongoing refinement, and inevitable oversight remains comparable to or higher than a skilled QA lead doing planning directly.
Cost vs. human wageclaude-sonnet-52/5While AI can draft test plan templates cheaply, the oversight, contextual judgment, and stakeholder negotiation needed keep the human cost dominant, making the all-in cost comparable rather than dramatically cheaper.
Technical feasibility todayclaude-haiku-4-5-202510012/5No mature products reliably handle full test schedule planning autonomously in production. AI tools can suggest test cases or timelines from templates, but real-world planning involves stakeholder negotiation, scope changes, and context-dependent trade-offs that require human oversight.
Technical feasibility todayclaude-sonnet-52/5AI-assisted test planning features exist in some ALM/QA tools but remain narrow, requiring significant human validation and adjustment before use in production planning.

Collaborate with field staff or customers to evaluate or diagnose problems and recommend possible solutions.

34

CI 30–38 · exposure 25 · augmentation 75 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of AI-driven troubleshooting assistants in QA teams remains in the pilot and early-adoption phase; most enterprises still rely heavily on human QA analysts for customer collaboration and diagnosis.
Sector adoption velocityclaude-sonnet-53/5Software/tech sectors are fast AI adopters overall, but this specific interpersonal diagnostic collaboration task sees more pilot-stage AI assistance (e.g., support ticket triage) than deep production automation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI demonstrates strong augmentation potential here: large language models can pre-screen tickets, suggest diagnoses, draft responses, and pull relevant documentation while QA analysts retain judgment on final recommendations and customer interaction—substantially improving throughput.
Augmentation potentialclaude-sonnet-54/5AI tools can significantly help by summarizing customer reports, suggesting likely root causes, and drafting solution recommendations, meaningfully speeding up the analyst's diagnostic workflow.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in basic troubleshooting and log analysis, this task fundamentally requires real-time interaction with field staff or customers to understand nuanced context, build trust, and adapt responses to complex, unpredictable problem scenarios—capabilities that remain immature in current AI systems.
Task automatabilityclaude-sonnet-52/5This task requires live, interpersonal collaboration and contextual diagnosis with humans, which current AI can support but not fully replace end-to-end without significant human involvement in gathering and interpreting nuanced context.
Adoption barriersclaude-haiku-4-5-202510013/5Customer-facing problem-solving involves some organizational friction (preference for human contact, trust concerns) but no hard legal or licensing barriers, though liability for incorrect diagnoses creates practical friction in many organizations.
Adoption barriersclaude-sonnet-52/5No formal licensing requirement, but customer trust, communication nuance, and organizational preference for human contact in escalations create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI-assisted troubleshooting platforms still require skilled QA staff for validation, oversight, and handling edge cases, meaning the all-in cost (inference + integration + human verification) remains comparable to or exceeds direct human diagnosis.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply draft responses or suggest diagnostics, but the human relationship-building, judgment, and ambiguous problem-solving still require costly QA analyst time, so overall cost savings are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5No mature production systems reliably handle the full scope of customer collaboration, diagnosis, and solution recommendation in real time; chatbots and ticketing integrations exist but introduce material error rates and require significant human oversight.
Technical feasibility todayclaude-sonnet-52/5AI chatbots and copilots exist to help triage bug reports or suggest fixes, but reliable autonomous collaboration with field staff/customers to diagnose novel problems is not yet a mature deployed product capability.

Recommend purchase of equipment to control dust, temperature, or humidity in area of system installation.

29

CI 20–39 · exposure 20 · augmentation 50 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5QA teams remain largely traditional in their approach to facility management; while some organizations use asset management and procurement software, AI-driven equipment recommendation for environmental control is not yet widely adopted in production across the sector.
Sector adoption velocityclaude-sonnet-52/5This is a narrow, infrequent task within software QA roles that isn't a focus of current enterprise AI adoption efforts, which are concentrated on core testing and coding tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist QA specialists by curating equipment options, comparing specifications, and generating initial recommendation drafts, but human judgment on facility-specific constraints, budget, and vendor relationships remains central to the task.
Augmentation potentialclaude-sonnet-53/5AI can help research equipment specifications, industry standards, and draft justification reports, offering moderate assistance to a human who still performs the site assessment and final call.
Task automatabilityclaude-haiku-4-5-202510012/5AI can gather information about environmental control equipment and suggest options based on specifications, but the task requires context-specific assessment of facility requirements, cost-benefit analysis, and vendor evaluation that are difficult to automate end-to-end without significant human oversight and domain expertise.
Task automatabilityclaude-sonnet-52/5This requires site-specific physical assessment and vendor/equipment recommendations tied to real-world environmental conditions, which AI cannot directly observe or verify; AI could draft generic recommendations but not perform the core judgment end-to-end.》 Limited automatability since much of the value is physical-site judgment.》,
Adoption barriersclaude-haiku-4-5-202510014/5Strong organizational and procedural barriers exist: purchasing decisions typically require sign-off from facilities or procurement specialists with legal accountability, budget authority is often vested in specific roles, and liability for equipment failure falls on authorized decision-makers rather than AI systems.
Adoption barriersclaude-sonnet-52/5No strict licensing requirement for this recommendation task, but organizational reliance on physical inspection and facilities expertise creates some friction against pure AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI tools for equipment research and recommendation generation are relatively inexpensive, but integration with procurement workflows and the need for human expertise review means total cost is roughly comparable to having a specialist perform initial research and recommendation analysis.
Cost vs. human wageclaude-sonnet-52/5AI could cheaply generate generic equipment suggestions, but a human still must inspect the site and validate specs, so overall cost savings versus a qualified analyst doing this occasional task are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI systems can search product databases and generate equipment recommendations, deployed products lack the facility-specific knowledge, vendor relationship context, and accountability required to make reliable purchasing recommendations that would satisfy organizational procurement standards.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously assesses installation environments and recommends specific HVAC/dust-control equipment purchases; this is a niche, low-frequency task with no commercial AI tooling targeting it.

Develop or specify standards, methods, or procedures to determine product quality or release readiness.

29

CI 25–32 · exposure 25 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While QA teams use AI-assisted testing tools, the strategic task of defining standards and procedures has low automation adoption. Most organizations still rely on human QA leadership to set quality policies, as this task requires judgment, accountability, and organizational credibility that teams are reluctant to delegate to AI.
Sector adoption velocityclaude-sonnet-53/5Software engineering and QA teams are moderately fast adopters of AI tooling (code review, test generation), but standard-setting and governance functions lag behind more tactical AI-assisted coding tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist QA analysts by generating template procedures, analyzing competitor or industry standards, or suggesting improvements to existing quality criteria. However, the human analyst must evaluate, customize, and validate the output, making this a helpful but not transformative augmentation.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully help draft quality standards, benchmark against industry practices, and synthesize past defect data into proposed criteria, significantly speeding up the analyst's initial drafting work.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in generating draft QA standards or procedures based on templates and best practices, developing or specifying standards requires human judgment about organizational context, risk tolerance, and strategic quality goals. AI cannot autonomously define what constitutes acceptable quality or release criteria for a product without substantial human oversight and validation.
Task automatabilityclaude-sonnet-52/5This requires organizational judgment, risk tolerance calibration, and stakeholder alignment that current AI cannot fully replace, though it can draft candidate standards or checklists. Most of the task involves decisions requiring context AI lacks.
Adoption barriersclaude-haiku-4-5-202510014/5QA standards and release procedures carry legal and liability weight; non-compliance can result in product failures, customer impact, and regulatory exposure. Organizations typically require human expertise and sign-off from senior QA leads or compliance officers, creating strong organizational and liability barriers to full automation.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement, but organizational accountability, liability for release decisions, and internal governance processes create meaningful friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for standard-generation assistance is cheap, but the task requires significant human review, validation, and refinement to produce usable organizational standards. The overhead of human oversight and rework makes the total cost comparable to or higher than having a QA analyst develop standards from scratch.
Cost vs. human wageclaude-sonnet-52/5Human QA leads and engineering managers must still interpret business risk and organizational context, so AI assistance reduces drafting time but doesn't replace the costly judgment-driven core work.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI product reliably develops QA standards or release procedures end-to-end. AI tools can draft documentation or suggest improvements to existing procedures, but they lack the domain expertise and organizational knowledge needed to establish authoritative quality standards that teams will trust and follow.
Technical feasibility todayclaude-sonnet-52/5Some AI tools assist in generating test plans or quality gates from requirements, but no deployed product autonomously defines organizational release-readiness standards reliably in production.

Participate in product design reviews to provide input on functional requirements, product designs, schedules, or potential problems.

29

CI 25–32 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Most software teams continue to rely on human QA analysts in design reviews; adoption of AI as a review participant remains at pilot or tool-assisted stage, not production replacement. Cultural inertia and the value of synchronous human judgment slow adoption.
Sector adoption velocityclaude-sonnet-53/5Software industry has moderate-to-fast AI adoption for coding and testing assistance, but review meetings involving cross-functional human judgment remain a slower-adopting subset of the workflow.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist a QA analyst by pre-analyzing designs, generating checklists of potential issues, or documenting requirements—useful preparation work—but the core task of informed participation in review meetings remains human-driven.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by pre-analyzing requirements documents, flagging ambiguities, generating test scenarios, and summarizing potential risks, boosting the analyst's preparation and contribution quality during reviews.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze design documents and flag obvious issues, it cannot reliably provide strategic input on product feasibility, organizational constraints, or trade-offs that require domain expertise and judgment. AI cannot fully participate in interactive review meetings where nuanced discussion and real-time decision-making occur.
Task automatabilityclaude-sonnet-52/5This task requires synthesizing cross-functional context, participating in live discussions, and exercising judgment about design tradeoffs, which current AI cannot fully replicate end-to-end despite being able to draft review notes or flag inconsistencies.dll
Adoption barriersclaude-haiku-4-5-202510014/5Design reviews are inherently collaborative and require trusted team participation; organizations strongly prefer human reviewers who take ownership, understand context, and participate in real-time discussion. Liability and accountability for design decisions rest with humans.
Adoption barriersclaude-sonnet-53/5There's no formal licensing requirement, but organizational norms mean design reviews are collaborative human meetings where accountability, communication, and stakeholder buy-in create meaningful friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5The overhead of integrating AI into design review workflows, training systems on product context, and managing output quality likely exceeds the cost of a QA analyst attending meetings. Human reviewers provide anchored institutional knowledge.
Cost vs. human wageclaude-sonnet-52/5Human QA analysts bring contextual and organizational knowledge that would require significant integration and oversight for AI to approximate, making the all-in cost of an AI substitute comparable to or higher than the marginal cost of an experienced human participant.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some tools exist to auto-review code and documentation for defects, but no deployed product reliably participates in collaborative design reviews as a full reviewer. AI systems struggle with context-dependent judgment about architectural decisions and requirements prioritization in real teams.
Technical feasibility todayclaude-sonnet-52/5Some AI tools can analyze specs or generate test-case suggestions for review, but no deployed product autonomously participates in design reviews as a stakeholder providing substantive judgment on schedules and tradeoffs.

Visit beta testing sites to evaluate software performance.

25

CI 13–38 · exposure 13 · augmentation 38 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Software QA has moderate AI adoption with automated test generation and cloud-based testing platforms becoming common, but on-site beta evaluation remains largely manual. Pilots of remote monitoring exist, but production displacement of site visits is slow.
Sector adoption velocityclaude-sonnet-52/5While software QA as a field sees moderate AI tool adoption for test case generation and defect detection, the specific on-site visit component sees essentially no AI adoption.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist QA analysts by pre-processing performance data, generating test reports, and flagging anomalies before or after site visits, improving efficiency in analysis and documentation, though the visit itself remains human-driven.
Augmentation potentialclaude-sonnet-52/5AI can help analyze data or logs collected during the visit and assist in drafting reports afterward, but it provides little to no assistance during the actual on-site evaluation itself.
Task automatabilityclaude-haiku-4-5-202510012/5Visiting physical beta testing sites involves in-person presence and contextual judgment about software behavior in real environments. While AI could analyze logs and performance metrics remotely, the core act of site visitation and real-time interactive evaluation remains human-dependent and cannot achieve the 50% time savings bar end-to-end.
Task automatabilityclaude-sonnet-51/5This task requires physical travel to a site and in-person observation/interaction, which current AI systems cannot perform; only the data-analysis portion afterward could be assisted.
Adoption barriersclaude-haiku-4-5-202510013/5Site visits may require security clearance, non-disclosure agreements, or client-site authorization; however, these are procedural rather than hard legal barriers. Organizations often prefer human testers on-site for relationship-building and real-time feedback.
Adoption barriersclaude-sonnet-52/5No licensing or legal requirement mandates a human specifically, but the inherent physical/on-site nature of the task creates a structural barrier to automation rather than a regulatory one.
Cost vs. human wageclaude-haiku-4-5-202510012/5The labor cost of a QA analyst making site visits is modest per hour ($25–50 loaded), and remote automation solutions still require significant integration and human oversight to validate findings, keeping all-in AI costs comparable or higher than manual visits.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical visit at all, so there is no viable cost comparison—human travel and on-site evaluation costs remain unavoidable.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed autonomous systems reliably visit physical locations and conduct interactive software testing without human oversight. Remote testing automation exists, but on-site evaluation requiring navigation, device interaction, and environmental context assessment has no mature production solution.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product can physically visit a site or conduct in-person beta evaluation; this remains entirely a human logistical and interpersonal activity.

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.