Petroleum Pump System Operators, Refinery Operators, and Gaugers

51-8093.00
Median wage $96,710/yr33,160 employed (US)Rank #478 of 923 scored · top 52% by substitution

Operate or control petroleum refining or processing units. May specialize in controlling manifold and pumping systems, gauging or testing oil in storage tanks, or regulating the flow of oil into pipelines.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution28
Exposure27
Augmentation51

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

24 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

8%

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%27

panel mean rating 2.1/5 → substitution pressure 27/100

Technical feasibility todayw 20%28

panel mean rating 2.1/5 → substitution pressure 28/100

Cost vs. human wagew 15%30

panel mean rating 2.2/5 → substitution pressure 30/100

Adoption barriersw 20%inverted — strong barriers lower the score28

panel mean rating 3.9/5 (barrier strength) → substitution pressure 28/100

Sector adoption velocityw 10%25

panel mean rating 2.0/5 → substitution pressure 25/100

Task breakdown (24 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.

Calculate test result values, using standard formulas.

76

CI 69–84 · exposure 80 · augmentation 75 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Refineries are highly digitized, capital-intensive operations with strong incentives to automate routine data processing. Process control systems and lab information management systems are already widespread in the sector, indicating rapid adoption of formula-automation tools.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a heavy industrial sector with slower digitization and automation adoption compared to information/professional services, though calculation-specific tools are common.
Augmentation potentialclaude-haiku-4-5-202510014/5AI and automated calculation systems significantly assist operators by instantly providing accurate results, reducing manual arithmetic errors and freeing cognitive capacity for interpretation and decision-making. This is already standard practice in refinery operations.
Augmentation potentialclaude-sonnet-54/5AI and automated calculation tools significantly speed up and reduce errors in performing standard formula-based test result calculations, letting operators focus on interpretation and decision-making.
Task automatabilityclaude-haiku-4-5-202510014/5Calculating test result values using standard formulas is a highly structured, deterministic task well-suited to automation. Modern AI systems and traditional software can reliably apply mathematical formulas to input data with minimal human intervention, easily achieving 50% time savings or better.
Task automatabilityclaude-sonnet-54/5Calculating test results from standard formulas is a deterministic, rule-based computation task well within current AI and even simpler software capabilities, given the input data.-
Adoption barriersclaude-haiku-4-5-202510013/5While the calculation itself is automatable, results often feed into regulated safety and quality documentation; regulatory requirements and operator oversight duties may require a human to review or validate outputs, creating moderate friction around full automation.
Adoption barriersclaude-sonnet-52/5Some regulatory documentation and sign-off requirements exist in refinery operations, but the calculation itself is not subject to licensing restrictions, only the broader gauging/certification process may be.
Cost vs. human wageclaude-haiku-4-5-202510015/5Once implemented, automated formula calculation costs almost nothing per execution (negligible compute and storage), whereas a human operator's loaded wage is substantial. The cost advantage is orders of magnitude.
Cost vs. human wageclaude-sonnet-55/5Automated calculation is computationally trivial and vastly cheaper than manual calculation by a skilled technician once formulas are digitized.
Technical feasibility todayclaude-haiku-4-5-202510015/5Deployed spreadsheet software, specialized lab information systems, and process control software routinely perform formula-based calculations at production scale in refineries today. This is a mature, reliable capability in real operations.
Technical feasibility todayclaude-sonnet-54/5Production systems (spreadsheets, LIMS software, SCADA-integrated calculators) already perform these standardized calculations reliably in refineries today, though full AI-driven autonomous handling of edge cases is less common.

Read automatic gauges at specified intervals to determine the flow rate of oil into or from tanks, and the amount of oil in tanks.

75

CI 71–79 · exposure 75 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Refinery and petroleum operations are capital-intensive, digitized sectors with strong incentives to automate routine monitoring; SCADA and automated instrumentation are already widespread industry practice, indicating rapid, deep adoption of gauge-reading automation.
Sector adoption velocityclaude-sonnet-53/5Oil and gas is a capital-intensive but somewhat conservative industry; large modern refineries have adopted automation extensively, but many facilities still rely on manual rounds due to legacy infrastructure.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-assisted dashboards and alert systems meaningfully improve operator situational awareness and decision-making, allowing human operators to focus on anomaly response rather than routine reading. However, this is primarily a support function rather than a transformative productivity multiplier.
Augmentation potentialclaude-sonnet-54/5AI-enabled monitoring systems can alert operators to anomalies, predict maintenance needs, and reduce the frequency of manual checks, significantly boosting operator efficiency while retaining oversight.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI vision systems can reliably read analog and digital gauges automatically, and SCADA/IoT integration enables continuous flow-rate monitoring without human intervention. This achieves >50% time saving by eliminating manual gauge checks, though occasional physical verification may still be required in practice.
Task automatabilityclaude-sonnet-54/5Reading automatic gauges at intervals is fundamentally a data-collection task that can be replaced by SCADA/telemetry systems feeding data directly to monitoring dashboards or automated logging systems.-Only physical verification or anomaly response requires a human.
Adoption barriersclaude-haiku-4-5-202510013/5Refineries face regulatory compliance and safety certification requirements that often mandate human attestation or sign-off on tank levels and flow rates, creating procedural friction. However, these are primarily oversight requirements rather than absolute prohibitions on automation itself.
Adoption barriersclaude-sonnet-52/5Some safety and regulatory inspection protocols may require periodic human verification of gauges, but there is no licensing requirement mandating a human specifically perform routine gauge readings.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated sensor systems and vision-based monitoring cost far less per reading than a human operator's loaded wage, especially across thousands of daily interval checks. Integration costs are low relative to elimination of repetitive manual labor.
Cost vs. human wageclaude-sonnet-55/5Automated sensors and telemetry systems have very low marginal cost per reading compared to a human operator's time and wage for periodic manual checks.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed computer vision products and industrial IoT monitoring systems already perform gauge reading and flow tracking in refinery environments at scale. While some manual oversight remains standard practice, the core automation is production-ready in major refineries.
Technical feasibility todayclaude-sonnet-54/5Refineries widely deploy SCADA, DCS, and remote telemetry systems that automatically log tank levels and flow rates without manual gauge reading, though some legacy sites still require manual checks.

Prepare calculations for receipts and deliveries of oil and oil products.

64

CI 60–67 · exposure 70 · augmentation 75 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Petroleum refining and distribution are capital-intensive, mature industries with slower digitization than software or finance sectors. While some large operators pilot automation, widespread production adoption of AI for calculation tasks remains limited, particularly at smaller facilities and independents.
Sector adoption velocityclaude-sonnet-53/5Oil and gas midstream/refining is a heavy industry with slower digitization than software/finance, though custody transfer automation and tank gauging systems have been adopted for years in larger operations.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can assist operators by rapidly checking calculations, flagging discrepancies, and auto-populating standard forms, substantially reducing error and time spent on reconciliation. The human operator retains critical oversight on complex shipments, regulatory compliance, and exceptions.
Augmentation potentialclaude-sonnet-54/5AI and automated calculation tools substantially speed up and reduce errors in volume/quality calculations, letting gaugers/operators focus on verification and exception handling.
Task automatabilityclaude-haiku-4-5-202510014/5Receipt and delivery calculations for oil products involve routine arithmetic, data entry, and reconciliation of quantities and specifications. Current AI systems with document processing and calculation capabilities can automate most of this workflow, though integration with legacy refinery systems and handling of non-standard formats may require some human oversight.
Task automatabilityclaude-sonnet-54/5Calculations for oil receipts/deliveries are structured, numeric tasks based on tank gauges, meters, and standard formulas—well-suited to software/AI automation with existing sensor data feeds.imit.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory requirements for inventory accounting, chain-of-custody documentation, and reconciliation create moderate friction. Many refineries and product terminals have compliance audits and legal liability tied to accurate reporting, which drives preference for human sign-off. However, the task itself is not legally restricted to a licensed human.
Adoption barriersclaude-sonnet-53/5Custody transfer for commercial oil transactions often requires certified metering and auditable, sometimes human-verified records for regulatory and contractual compliance, creating moderate friction even though the calculation itself is straightforward.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs for automated calculation and verification are substantially lower than the fully loaded wage of a petroleum pump operator or gauger performing these tasks manually. Overhead per task-equivalent is minor once the system is integrated.
Cost vs. human wageclaude-sonnet-54/5Automated calculation systems are cheap to run once integrated, versus paying an operator's time to manually compute volumes, though integration with legacy SCADA/metering systems adds some cost.
Technical feasibility todayclaude-haiku-4-5-202510013/5Document processing AI and accounting automation tools exist and are deployed in some energy operations, but many refineries still rely on legacy systems and manual validation. Production systems handle standard receipts reliably, but performance degrades with unusual shipments, damaged documentation, or non-standard procedures.
Technical feasibility todayclaude-sonnet-54/5Tank gauging and custody transfer calculation software is already deployed widely in refineries and terminals, performing these calculations automatically from instrumentation data.

Plan movement of products through lines to processing, storage, and shipping units, using knowledge of system interconnections and capacities.

52

CI 25–79 · exposure 58 · augmentation 88 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Petroleum refining and chemical processing are highly digitized, capital-intensive sectors with strong incentives to optimize throughput and safety. Adoption of AI-assisted and automated planning is already visible in major integrated operations.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a slower-adopting, capital-intensive physical industry where AI is used for optimization pilots but full autonomous planning is not widespread in production.
Augmentation potentialclaude-haiku-4-5-202510015/5AI assists human operators by rapidly generating feasible plans, flagging constraints, and simulating outcomes, substantially raising the speed and quality of human decision-making while the operator retains final authority and oversight.
Augmentation potentialclaude-sonnet-54/5AI-based scheduling and optimization tools can meaningfully assist operators in planning product movement by simulating scenarios and flagging capacity constraints, improving efficiency while humans remain accountable for final decisions.
Task automatabilityclaude-haiku-4-5-202510015/5AI systems can analyze system interconnections, capacities, flow rates, and product specifications to generate optimal routing plans and schedules. Industrial optimization and constraint-satisfaction tools exist off-the-shelf and can reduce planning time by >50% while meeting safety and efficiency standards.
Task automatabilityclaude-sonnet-52/5Planning product flow requires real-time integration of tank capacities, line schedules, safety constraints, and physical system state that current AI cannot fully manage autonomously without extensive custom modeling.dec Optimization software exists but a full end-to-end AI takeover of dynamic planning with equal quality is not yet demonstrated broadly.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory oversight, safety certification requirements, and industry custom generally require licensed operators to validate and approve AI-generated plans before execution. Liability and operational risk create friction but do not completely prevent automation.
Adoption barriersclaude-sonnet-54/5Refinery operations are heavily regulated for safety (OSHA, EPA) and typically require certified operators to oversee product movement due to high liability and physical risk of error, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5Industrial optimization software and agent systems are expensive to deploy and maintain, but their cost is now typically lower than employing dedicated human planners for routine scheduling, especially across multi-unit operations.
Cost vs. human wageclaude-sonnet-52/5Refinery-grade optimization/scheduling software and the integration, sensors, and safety validation needed are costly to deploy and maintain, and skilled operator oversight remains necessary, keeping cost savings modest relative to labor.
Technical feasibility todayclaude-haiku-4-5-202510014/5Advanced process control and scheduling software is deployed in production at major refineries and chemical plants, integrating real-time system data and constraints. However, most systems still require human review for final approval due to safety criticality and edge cases.
Technical feasibility todayclaude-sonnet-52/5Advanced process control and scheduling optimization tools exist in refineries, but they are decision-support systems requiring operator interpretation and adjustment, not autonomous planning agents performing this task reliably unsupervised.

Record and compile operating data, instrument readings, documentation, and results of laboratory analyses.

50

CI 28–72 · exposure 50 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Petroleum refining has moderate AI adoption for predictive maintenance and process optimization, but manual data recording and compilation remain standard practice. Some facilities pilot automated logging, but sector-wide displacement is slow due to regulatory lock-in and operator-certification requirements.
Sector adoption velocityclaude-sonnet-53/5Oil and gas/refining is a capital-intensive, moderately digitized sector with established SCADA/historian adoption but slower uptake of newer AI-driven data compilation tools compared to information-sector industries.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist operators by auto-populating structured fields, flagging anomalous readings, organizing lab results, and generating summary reports, meaningfully reducing transcription and compilation time while the operator retains final verification and sign-off responsibility.
Augmentation potentialclaude-sonnet-54/5AI and automated systems significantly reduce manual data entry burden and enable operators to focus on interpretation and exception-handling rather than transcription.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can extract and organize structured instrument readings and lab results, the task requires real-time monitoring of complex systems, context-dependent interpretation, and integration with safety protocols that demand human judgment today. Current systems lack the robustness needed to fully replace operators' data compilation in safety-critical refinery settings.
Task automatabilityclaude-sonnet-54/5Recording and compiling structured operating data, instrument readings, and lab results is largely a data-transcription and aggregation task well-suited to automation via sensors, SCADA/DCS historians, and software integration, though some manual verification remains.
Adoption barriersclaude-haiku-4-5-202510014/5Refineries operate under strict EPA, OSHA, and industry-specific regulations (API standards) requiring documented accountability and chain-of-custody for safety-critical readings. Human operators must legally certify and sign-off on records, creating a hard compliance barrier to full automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement mandates a human record data, but safety-critical process environments impose some oversight and validation requirements before figures are relied upon for compliance or safety decisions.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of AI systems into refinery SCADA/documentation infrastructure involves substantial engineering and maintenance costs. Compared to a single operator's annual wage, full automation would require capital investment and ongoing oversight that approaches or exceeds the human cost in most refinery operations.
Cost vs. human wageclaude-sonnet-54/5Automated data logging and compilation software has low marginal cost per data point once integrated, far cheaper than continuous manual recording by an operator, though integration and system costs are non-trivial upfront.
Technical feasibility todayclaude-haiku-4-5-202510012/5Automated data logging systems exist for some refinery parameters, but end-to-end compilation integrating instrument readings, lab results, and documentation remains largely manual or semi-automated with significant human oversight required. No mature AI product reliably performs this full task in production refinery environments without operator verification.
Technical feasibility todayclaude-sonnet-54/5Data historians, SCADA systems, and LIMS (lab information management systems) already automate much of this recording and compilation in production refineries today, though full end-to-end automation across all data sources varies by site.

Monitor process indicators, instruments, gauges, and meters to detect and report any possible problems.

34

CI 25–43 · exposure 38 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refineries and petrochemical facilities are adopting monitoring automation incrementally (SCADA, digital twins, predictive maintenance), but operator presence and human decision-making remain mandated by regulation; adoption is slower than in information-intensive sectors due to safety and liability constraints.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, safety-conservative industry with slower digitization and AI adoption compared to information/finance sectors, though predictive maintenance pilots are growing.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-driven dashboards, real-time anomaly alerts, and predictive analytics can assist operators in interpreting complex multi-parameter data streams and reduce cognitive load, moderately improving their situational awareness and response time, though the core task remains human-centered.
Augmentation potentialclaude-sonnet-54/5AI-based anomaly detection and predictive analytics significantly enhance operators' ability to catch issues early and prioritize attention, while humans remain responsible for final judgment and intervention.
Task automatabilityclaude-haiku-4-5-202510012/5Monitoring gauges and meters can be partially automated with sensor integration and anomaly detection algorithms, but end-to-end automation meeting the 50% time-saving threshold is limited by the need for human judgment in interpreting complex sensor interactions, environmental context, and responding to unexpected conditions in a safety-critical refinery setting.
Task automatabilityclaude-sonnet-53/5Modern SCADA/DCS systems with anomaly detection can automate much of the monitoring and alerting, but physical gauging, sensor validation, and edge-case judgment still require human presence and interpretation on-site.
Adoption barriersclaude-haiku-4-5-202510014/5Refineries operate under strict EPA, OSHA, and API regulations that mandate trained, licensed operators responsible for safety and environmental compliance; liability for automation-related safety incidents is asymmetric and heavily regulated, creating legal and certification barriers to autonomous operation.
Adoption barriersclaude-sonnet-54/5Process safety regulations (OSHA PSM, EPA) and liability for catastrophic failures (explosions, spills) mean human oversight and sign-off are typically mandated, creating strong regulatory and safety barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5Industrial monitoring systems (sensors, servers, integration) have significant upfront and maintenance costs, and continuous human oversight remains necessary in refineries; the all-in cost (hardware, software, integration, liability, redundancy) approaches or exceeds the loaded cost of a skilled operator.
Cost vs. human wageclaude-sonnet-53/5Sensor networks and monitoring software have high upfront integration and maintenance costs comparable to operator wages when factoring in redundancy and safety validation, though incremental software costs are low once installed.
Technical feasibility todayclaude-haiku-4-5-202510012/5While industrial monitoring systems and SCADA dashboards with alerts exist in production, they typically flag anomalies rather than independently detect and report all problems; systems require human operators to interpret and act, and fully autonomous problem detection in complex refinery environments remains research-heavy with material error rates.
Technical feasibility todayclaude-sonnet-53/5Industrial control systems with automated alarms and predictive analytics are deployed in refineries today, but full autonomous monitoring without human operators is not standard due to safety-critical requirements and legacy equipment variability.

Start pumps and open valves or use automated equipment to regulate the flow of oil in pipelines and into and out of tanks.

32

CI 25–40 · exposure 30 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Refineries have deployed SCADA and process control automation for decades, but human operators remain in supervisory roles and cannot be fully removed due to safety and regulatory constraints. Adoption of full autonomy is slower than in information-sector tasks, reflecting the physical, safety-critical nature of the work.
Sector adoption velocityclaude-sonnet-52/5Oil and gas is a capital-intensive, safety-critical industrial sector with slower AI adoption relative to information/professional services, though SCADA/automation adoption has been ongoing for decades.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-driven monitoring systems, predictive analytics, and automated alerts significantly enhance operator productivity by flagging anomalies and optimizing flow rates in real time, allowing operators to manage larger, more complex systems while maintaining safety oversight.
Augmentation potentialclaude-sonnet-53/5AI-enhanced predictive analytics and anomaly detection can help operators monitor flow rates and predict equipment failures, improving decision-making while humans remain responsible for physical operation and safety compliance.
Task automatabilityclaude-haiku-4-5-202510012/5While opening valves and starting pumps can be automated via SCADA systems, current AI cannot reliably assess the dynamic operational context (pressure anomalies, safety interlocks, emergency conditions) that requires human judgment. The task requires situational awareness and immediate physical response to equipment that AI systems do not possess.
Task automatabilityclaude-sonnet-52/5This is a physical/control-room task requiring hands-on valve operation and pump activation in hazardous industrial environments; while SCADA systems already automate much of the routine flow regulation, full end-to-end automation including physical intervention and anomaly response is not achievable with off-the-shelf AI today.5
Adoption barriersclaude-haiku-4-5-202510014/5Petroleum refining is heavily regulated (EPA, OSHA, PSM regulations); operators must be licensed and are legally responsible for safe operation. Liability and safety-critical certification requirements create substantial barriers to full automation, and regulatory frameworks mandate human accountability in hazardous operations.
Adoption barriersclaude-sonnet-54/5Petroleum operations are heavily regulated (OSHA, EPA, process safety management) with liability concerns for spills/explosions, requiring certified/licensed operators to be present and accountable for pipeline and tank operations.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated SCADA systems have very low ongoing operational cost compared to a fully-loaded petroleum operator wage (~$60k–80k/year). Once deployed, the marginal cost per operation is negligible, making AI-driven automation far cheaper per unit task.
Cost vs. human wageclaude-sonnet-52/5Existing automated control systems already capture much of the cost savings; incremental AI layered on top offers marginal additional savings while requiring costly integration with legacy industrial equipment and safety certification.
Technical feasibility todayclaude-haiku-4-5-202510013/5SCADA and industrial control systems exist and perform routine valve/pump operations in refineries, but they operate within pre-programmed parameters and require human operators for anomaly detection, manual overrides, and safety decisions. These systems are not fully autonomous AI agents but rather rule-based automation with human oversight.
Technical feasibility todayclaude-sonnet-52/5Industrial control systems (DCS/SCADA) with programmed automation exist and are deployed, but these are decades-old automation technologies rather than AI in the modern sense, and human operators remain required for oversight, exceptions, and physical valve/pump actions.

Signal other workers by telephone or radio to operate pumps, open and close valves, and check temperatures.

25

CI 25–25 · exposure 25 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refinery operations are capital-intensive and risk-averse; adoption of AI for autonomous worker coordination is slow, with most facilities still relying on established radio/telephone protocols and human judgment rather than AI dispatch systems.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, safety-regulated sector with slower digitization and AI adoption compared to information/professional services.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-driven monitoring dashboards and automated alerts can assist operators by surfacing temperature/pressure anomalies and recommending actions, raising situational awareness without removing the human from the decision loop.
Augmentation potentialclaude-sonnet-53/5AI-enabled monitoring dashboards, predictive alerts, and automated logging can assist operators in tracking temperatures and coordinating tasks, improving efficiency while humans remain responsible for control actions.
Task automatabilityclaude-haiku-4-5-202510012/5AI systems can monitor sensors and generate alerts, but signaling human workers to perform physical operations requires real-time coordination in a safety-critical environment where judgment about timing and conditions cannot be fully automated without human oversight.
Task automatabilityclaude-sonnet-52/5This task involves real-time physical plant coordination and verbal communication tied to on-site sensory judgment (checking temperatures, valve states); AI can support monitoring but cannot fully replace the human coordination loop with equal quality end-to-end today.itution.
Adoption barriersclaude-haiku-4-5-202510014/5Refinery operations are heavily regulated (OSHA, EPA, API standards), and liability concerns mean a qualified human operator must remain responsible for safety-critical decisions; replacing human communication with autonomous signals faces legal and certification barriers.
Adoption barriersclaude-sonnet-54/5Safety-critical industrial processes involving valves, pressures, and temperatures are heavily regulated (OSHA, PSM) and typically require certified human oversight to prevent catastrophic errors.
Cost vs. human wageclaude-haiku-4-5-202510012/5Implementing automated signaling and monitoring systems requires significant infrastructure investment in sensors, networks, and integration; the cost per task execution likely exceeds the wage of a pump operator performing this coordination.
Cost vs. human wageclaude-sonnet-52/5Automating this requires significant sensor integration, control system upgrades, and safety validation, so all-in AI costs are not clearly cheaper than existing human operators for this narrow communication task.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can send alerts and communicate information, production systems for autonomous coordination of refinery operations with human workers remain immature; deployed solutions focus on monitoring, not autonomous dispatch of workers.
Technical feasibility todayclaude-sonnet-52/5Some SCADA/automation systems exist that reduce need for manual signaling, but deployed products don't autonomously perform the full communicative coordination task reliably across diverse refinery operations.

Verify that incoming and outgoing products are moving through the correct meters, and that meters are working properly.

25

CI 25–25 · exposure 25 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refineries are capital-intensive, risk-averse, and operate under strict regulatory oversight. Adoption of AI agents for critical flow-verification tasks remains minimal; monitoring dashboards exist, but real-time autonomous substitution of human verification is rare and slow-moving.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a capital-intensive, safety-conscious sector with slower digitization and AI adoption compared to information or financial services, though some predictive maintenance analytics are being piloted.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by continuously logging meter readings, visualizing trends, and alerting operators to anomalies, which raises situational awareness and reduces manual log review. However, the human operator remains the decision-maker for physical verification and corrective action.
Augmentation potentialclaude-sonnet-53/5AI-based anomaly detection and sensor analytics can flag meter discrepancies or unusual flow patterns, helping operators prioritize checks, though the physical verification and judgment remain human-led.
Task automatabilityclaude-haiku-4-5-202510012/5While AI systems could monitor meter readings and flag discrepancies, verifying physical correct-path flow and assessing meter mechanical function require on-site inspection and troubleshooting that current AI cannot reliably perform end-to-end. Partial automation of data logging and anomaly detection is possible, but does not meet the ≥50% time-saving bar for the full verification task.
Task automatabilityclaude-sonnet-52/5Physical verification of meter operation and product routing requires on-site sensing, valve/line tracing, and physical inspection that current AI cannot fully execute end-to-end; software can monitor sensor data but not perform the full physical verification task.
Adoption barriersclaude-haiku-4-5-202510014/5Refinery operations are heavily regulated (EPA, OSHA, industry safety standards); product custody and inventory control have strict compliance and liability requirements. Manual verification by licensed personnel is mandated in many jurisdictions, and automation substitution faces regulatory gatekeeping and fiduciary risk.
Adoption barriersclaude-sonnet-54/5Safety-critical infrastructure in refineries is heavily regulated (OSHA, environmental, process safety management), often requiring certified operators to verify and sign off on flow and metering integrity.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted monitoring (sensor logs, automated alerts) has low inference cost, but the overhead of human verification, on-site inspection, and maintenance still dominates total cost, and does not undercut the loaded labor cost of a skilled pump operator or gauger.
Cost vs. human wageclaude-sonnet-52/5Sensor and monitoring software is relatively cheap to run, but integration with legacy industrial control systems and required human oversight for safety-critical verification keeps effective cost comparable to or only modestly below human labor.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some industrial IoT platforms log meter readings and alert on out-of-range values, but reliably diagnosing whether meters are functioning correctly or products are routed to the wrong system requires human technician presence and judgment that products do not yet automate at scale in production refineries.
Technical feasibility todayclaude-sonnet-52/5SCADA/DCS systems with alarms and analytics exist and are deployed for flow monitoring, but full autonomous verification of correct routing and meter integrity without human confirmation is not standard practice.

Synchronize activities with other pumphouses to ensure a continuous flow of products and a minimum of contamination between products.

25

CI 25–25 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption in refining is slow; the sector prioritizes safety, regulatory compliance, and human expertise, and most sites maintain traditional operator-centric control with incremental digitization (SCADA upgrades) rather than agent-based automation.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, safety-conscious industry with slower digitization and AI adoption compared to sectors like finance or software.
Augmentation potentialclaude-haiku-4-5-202510013/5AI systems can usefully augment operators by monitoring multiple pumphouse feeds, predicting product transitions, and alerting to contamination risk, raising situational awareness and reducing manual monitoring workload while the operator retains control.
Augmentation potentialclaude-sonnet-53/5AI-based predictive analytics and control system recommendations can help operators anticipate flow issues and contamination risks, improving decision-making while humans remain responsible for execution.
Task automatabilityclaude-haiku-4-5-202510012/5This task requires real-time coordination across multiple physical systems and real-time decision-making based on process states. While AI can monitor sensor data and suggest adjustments, the distributed coordination across independent pumphouses and the need for human judgment during anomalies make end-to-end automation with ≥50% time savings infeasible with current systems.
Task automatabilityclaude-sonnet-52/5This requires real-time coordination with physical equipment, sensor monitoring, and physical/procedural safety judgment across facilities, which current general AI cannot execute end-to-end without deep integration into control systems.'
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: operator licenses and certifications are legally required in many jurisdictions, regulatory requirements (EPA, API standards) mandate human accountability for product quality and spill prevention, and liability for contamination or safety failures creates strong organizational and legal pressure to retain human sign-off.
Adoption barriersclaude-sonnet-54/5Safety-critical process operations in refineries are heavily regulated (OSHA PSM, EPA), requiring qualified human operators for oversight and liability reasons, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Deploying AI monitoring and coordination across multiple pumphouses with requisite integration, oversight, and safety redundancy would be expensive relative to a single operator's loaded wage, especially given the critical safety and contamination-prevention role.
Cost vs. human wageclaude-sonnet-52/5Implementing automated cross-facility synchronization requires expensive control system integration, sensors, and safety validation, making near-term AI cost comparable to or higher than a trained operator's wage.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably automates cross-pumphouse synchronization at scale; existing SCADA and DCS systems provide monitoring and alerts but require human operators to coordinate and execute decisions. Products can assist but do not operate the task autonomously in production environments.
Technical feasibility todayclaude-sonnet-52/5Some SCADA/DCS systems have automated coordination logic, but full autonomous synchronization across pumphouses to prevent contamination is still largely human-supervised in production refineries today.'

Operate control panels to coordinate and regulate process variables such as temperature and pressure, and to direct product flow rate, according to process schedules.

25

CI 20–30 · exposure 30 · augmentation 75 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refineries are conservative, capital-intensive operations with long asset lifecycles and safety-first cultures; adoption of AI-driven autonomous control is measured and limited to pilot projects and advisory layers rather than production displacement of operators.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a capital-intensive, safety-conservative physical industry with slower AI adoption compared to information-sector benchmarks, though APC and predictive analytics tools are gradually being adopted.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered dashboards, predictive analytics, and anomaly detection meaningfully assist operators by synthesizing large data streams, forecasting process deviations, and recommending adjustments, thereby raising their real-time decision quality and efficiency while they remain in full control.
Augmentation potentialclaude-sonnet-54/5AI-driven APC, predictive analytics, and anomaly detection significantly assist operators in maintaining optimal process variables and catching deviations, meaningfully boosting productivity while humans remain in control.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could theoretically monitor parameters and suggest adjustments, coordinating multiple process variables in real-time refineries requires immediate human decision-making in dynamic, safety-critical environments where latency and error costs are prohibitive. Current AI monitoring systems are deployed only in advisory roles, not autonomous control.
Task automatabilityclaude-sonnet-52/5While DCS/SCADA systems already automate much routine setpoint control, safe operation of a refinery control panel requires continuous real-time judgment, anomaly response, and physical-world integration that current AI cannot fully replace end-to-end without significant human oversight.
Adoption barriersclaude-haiku-4-5-202510015/5Refinery operations are heavily regulated by EPA, OSHA, and industry safety standards that mandate licensed human operators and responsibility for critical safety decisions; liability for system failures rests on the responsible human, creating a hard legal barrier to full substitution.
Adoption barriersclaude-sonnet-54/5Refineries are highly regulated (OSHA, EPA, PSM requirements) and carry catastrophic liability/safety risk, so licensed, trained human operators are generally required to be present and accountable for control panel decisions.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI monitoring and advisory systems add cost (licensing, integration, human oversight) without eliminating the operator role, making the all-in cost comparable to or higher than human operator wages for the foreseeable future.
Cost vs. human wageclaude-sonnet-52/5APC systems reduce labor needs but require substantial capital investment, integration, and ongoing engineering support, making the all-in cost comparable to or only modestly cheaper than skilled operator labor rather than an order of magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed products can monitor refinery processes and flag anomalies, but fully autonomous operation of control panels regulating temperature, pressure, and flow across complex interdependent systems does not exist in production. Existing SCADA systems are rule-based, not AI-driven end-to-end automation.
Technical feasibility todayclaude-sonnet-53/5Advanced process control (APC) and automated control loops are mature and widely deployed, but full autonomous panel operation without a human operator overseeing safety-critical decisions is not a deployed reality.

Perform tests to check the qualities and grades of products, such as assessing levels of bottom sediment, water, and foreign materials in oil samples, using centrifugal testers.

24

CI 23–25 · exposure 25 · augmentation 38 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refinery operations remain highly physical and manually intensive with conservative automation practices. Adoption of AI in this specific laboratory testing domain is minimal, with most facilities still relying on trained human gaugers and traditional testing protocols.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, physically-oriented industrial sector with historically slower AI adoption compared to information-based industries, though some sensor automation exists.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist by automating result logging or flagging anomalies in data streams, but the task inherently requires physical sample handling and visual/sensory judgment. Augmentation potential is limited compared to cognitive office tasks.
Augmentation potentialclaude-sonnet-53/5AI-enabled sensors and data analytics can help operators track trends, flag anomalies, and streamline reporting of test results, improving efficiency even though the physical testing itself remains manual.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could analyze centrifuge readouts or interpret test results digitally, the physical task of obtaining oil samples, loading them into centrifugal testers, and observing real-time results requires hands-on laboratory work. Current AI systems cannot physically operate equipment or reliably extract representative samples without human intervention.
Task automatabilityclaude-sonnet-52/5This task requires physical sample collection and operation of centrifugal testing equipment, which current AI cannot perform end-to-end; only data interpretation portions could be assisted., leaving most of the hands-on work unautomated.
Adoption barriersclaude-haiku-4-5-202510014/5Oil quality grading affects product specifications, safety, and regulatory compliance; most refineries have established protocols requiring documented human verification and accountability. Liability concerns around misgraded products and regulatory requirements for certified technician sign-off create substantial adoption friction.
Adoption barriersclaude-sonnet-54/5Refinery operations are heavily regulated for safety and quality assurance, often requiring certified personnel to verify test results and sign off on product quality, creating strong institutional and liability barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI systems capable of automated testing would require significant hardware investment, integration with lab equipment, and continuous human oversight to ensure sample validity. The loaded cost per test would likely exceed the cost of a technician performing the manual task.
Cost vs. human wageclaude-sonnet-52/5Specialized automated testing equipment requires significant capital investment and maintenance, and human operators remain cost-competitive for this narrow physical task compared to bespoke automation systems.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI image recognition could potentially assist in reading meter outputs or analyzing visual test results, but no deployed product reliably performs the full workflow of sample preparation, centrifuge operation, and quality grading in production refinery settings. Existing solutions are limited to narrow aspects like result interpretation.
Technical feasibility todayclaude-sonnet-52/5While automated lab analyzers and sensor-based systems exist in refineries, fully autonomous physical sampling and centrifugal testing without human intervention is not yet standard deployed practice.

Operate auxiliary equipment and control multiple processing units during distilling or treating operations, moving controls that regulate valves, pumps, compressors, and auxiliary equipment.

21

CI 18–25 · exposure 25 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refining is a mature, safety-conscious, capital-intensive sector with strong institutional resistance to autonomous control of critical equipment. Adoption of AI for active process control remains extremely slow; most innovation remains in monitoring and decision-support rather than direct physical automation.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, safety-conscious, slow-adopting sector for new AI-driven control paradigms, with automation upgrades occurring incrementally over long equipment lifecycles.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could usefully augment operators by providing real-time sensor analysis, predictive alerts on equipment performance, and decision-support (e.g., recommended valve positions), helping them manage multiple units more efficiently while they retain full manual control and accountability.
Augmentation potentialclaude-sonnet-54/5Modern DCS, predictive analytics, and APC software significantly aid operators in monitoring and optimizing multiple units, improving efficiency and safety while keeping humans in the loop.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could theoretically monitor sensor data and suggest valve adjustments, the task requires real-time physical control of equipment (moving physical controls) and dynamic response to multiple simultaneous processing units in a safety-critical environment. Current AI cannot reliably manipulate physical control systems end-to-end with the redundancy and fail-safe oversight that petroleum refining demands.
Task automatabilityclaude-sonnet-52/5This involves physical control room and field operation of hazardous industrial equipment; while DCS/SCADA systems already automate routine control, the task as described requires continuous real-time judgment and physical presence that current AI cannot fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Petroleum refining is heavily regulated (EPA, OSHA, API standards); operators must be licensed and certified. Legal and liability frameworks require a licensed human to control critical processing units, and safety-critical equipment operation cannot be fully automated without extensive regulatory approval and fail-safe mechanisms.
Adoption barriersclaude-sonnet-54/5Process safety regulations (OSHA PSM, EPA RMP) and liability concerns for hazardous processing require certified human operators to be present and accountable, creating strong regulatory and safety barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integrating AI into safety-critical refinery control requires extensive validation, certification, redundancy, and human oversight infrastructure—often more expensive than the loaded wage of experienced operators who perform the task reliably and bear accountability.
Cost vs. human wageclaude-sonnet-52/5Control automation systems are costly to install, validate, and maintain in safety-critical refinery environments, and human operators remain necessary for oversight, so cost savings are limited relative to labor cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably operates multi-unit refinery control systems autonomously today. Industrial control systems exist (SCADA, DCS) but these are fixed automation, not AI agents. Regulatory and safety requirements mean any AI involvement in actual valve/pump control remains supervisory and human-supervised, not independent operation.
Technical feasibility todayclaude-sonnet-52/5Advanced process control (APC) and DCS systems are deployed widely, but they augment rather than replace the operator role described here, and full autonomous operation of multiple units without human oversight is not in production.

Read and analyze specifications, schedules, logs, test results, and laboratory recommendations to determine how to set equipment controls to produce the required qualities and quantities of products.

21

CI 18–25 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Petroleum refining remains a heavily manual, safety-critical, and regulated sector with slow AI adoption; while digitization is increasing, actual autonomous AI control of refinery equipment is not happening at scale in production today.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a capital-intensive, physically-oriented sector with historically slower AI adoption compared to information-based industries, though APC and predictive analytics pilots are growing.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automatically extracting and summarizing data from logs, specifications, and test results, and by flagging deviations or recommending control adjustments for human review; this moderately increases operator productivity but the human remains essential for final judgment and safety.
Augmentation potentialclaude-sonnet-54/5AI-based analytics and pattern recognition can meaningfully assist operators in synthesizing logs, test results, and specifications faster, improving decision speed while the operator remains responsible for final control settings.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can read and parse documents (specifications, logs, test results), determining precise equipment control settings requires real-time process knowledge, safety-critical judgment, and integration with live system feedback that current AI systems cannot reliably perform end-to-end without human intervention.
Task automatabilityclaude-sonnet-52/5While AI can help parse and analyze structured logs and specifications, translating this into safe, real-time equipment control settings on physical refinery hardware requires integration with proprietary control systems and physical verification that current off-the-shelf AI cannot fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Refinery operations are heavily regulated by EPA, OSHA, and industry standards; a licensed operator must legally supervise and sign off on equipment control changes, and liability for product quality and safety failures rests on the operator, creating hard legal and regulatory barriers to full automation.
Adoption barriersclaude-sonnet-54/5Refinery operations are subject to strict safety regulations, environmental compliance, and liability concerns that typically require a certified human operator to review and authorize control setting changes.
Cost vs. human wageclaude-haiku-4-5-202510012/5An AI document-reading assistant costs far less than a petroleum operator, but the task is not automatable end-to-end, so the comparison is misleading; full automation would need expensive integration with refinery control systems, making it comparable to or more expensive than human oversight.
Cost vs. human wageclaude-sonnet-52/5Deploying AI systems for this task requires expensive plant-specific integration, sensor infrastructure, and safety validation, making the all-in cost comparable to or higher than skilled operator labor in most facilities.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed production systems reliably autonomously set refinery equipment controls based on specifications and test data; existing process control systems are rule-based and require human operators to interpret data and make final control decisions.
Technical feasibility todayclaude-sonnet-52/5Advanced process control (APC) and some AI-assisted optimization tools exist in refineries, but they are narrow, heavily customized, and still require human operators to interpret and validate before acting on recommendations.

Lower thermometers into tanks to obtain temperature readings.

21

CI 7–35 · exposure 13 · augmentation 13 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refining is capital-intensive but conservative in automation adoption for manual field tasks; most refineries still rely on operator rounds and manual thermometer reading despite the industry's overall digitization, reflecting organizational inertia and safety-compliance friction.
Sector adoption velocityclaude-sonnet-52/5Refinery and petroleum operations are capital-intensive but adopt automation via fixed sensors/IoT rather than AI agents; this specific manual task shows slow AI-driven displacement, though automated temperature sensors have long existed as non-AI solutions.
Augmentation potentialclaude-haiku-4-5-202510012/5Digital thermometers and wireless probes can assist by reducing transcription errors and enabling remote monitoring, but AI offers minimal augmentation to the core physical act of inserting a thermometer and reading it.
Augmentation potentialclaude-sonnet-51/5AI does not meaningfully assist in the physical act of lowering a thermometer; any assistance would come from sensor automation, not AI augmentation of the human performing this manual task.
Task automatabilityclaude-haiku-4-5-202510012/5While temperature sensors could be automated, this task involves physical insertion of thermometers into tanks, requiring manual dexterity and real-time decision-making about insertion depth and safety protocols. Current AI cannot reliably perform the physical manipulation end-to-end with equivalent speed and safety.
Task automatabilityclaude-sonnet-51/5This is a physical manipulation task requiring lowering a physical instrument into a tank; current AI systems cannot perform physical manipulation tasks, and this would require robotics not language/vision AI., ratherthan being reduced to information processing.
Adoption barriersclaude-haiku-4-5-202510014/5Refineries operate under strict EPA and OSHA regulations requiring documented temperature monitoring for safety and compliance; workers must physically verify readings and sign logs, creating legal liability and audit requirements that resist full automation.
Adoption barriersclaude-sonnet-52/5No licensing barrier per se, but physical presence, safety protocols in refinery environments, and equipment reliability requirements create some friction against automation via non-purpose-built AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5Automating thermometer insertion would require custom robotics infrastructure (actuators, vision systems, tank-specific calibration) costing far more than the loaded wage of a pump operator performing this routine task multiple times daily.
Cost vs. human wageclaude-sonnet-51/5There is no AI system that replaces this physical task; the comparison would require robotic hardware, which is far more costly than a human performing this simple manual action.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI systems independently perform manual thermometer insertion in refinery tanks. Some refineries use automated temperature probes, but these are fixed installations, not AI agents that learn to insert traditional thermometers—this remains largely a manual task in production.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical thermometer insertion into tanks; this remains a manual or fixed-sensor-based operation, not an AI capability.

Patrol units to monitor the amount of oil in storage tanks, and to verify that activities and operations are safe, efficient, and in compliance with regulations.

21

CI 16–25 · exposure 17 · augmentation 50 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refineries are capital-intensive, risk-averse facilities operating under strict legacy constraints. While IoT sensor deployment has increased, most still rely on human patrollers as primary safeguards. Adoption of autonomous patrol systems remains limited to pilots at very large facilities; the sector overall lags in automation velocity.
Sector adoption velocityclaude-sonnet-52/5Oil and gas is a heavy industry with lower digitization rates and slower AI adoption compared to information/professional services sectors, though remote monitoring tech is increasingly used.
Augmentation potentialclaude-haiku-4-5-202510013/5Remote monitoring dashboards and automated alerts for tank levels and pressure anomalies meaningfully assist patrol operators in prioritizing inspections and reducing routine manual checks. However, these tools support rather than transform the task, since physical presence and judgment remain essential.
Augmentation potentialclaude-sonnet-53/5AI-enabled sensors, predictive analytics, and monitoring dashboards can assist operators by flagging anomalies and providing real-time data, improving efficiency without replacing the human patrol role.
Task automatabilityclaude-haiku-4-5-202510012/5While remote monitoring systems and sensors can track oil levels and some operational parameters, the task requires physical patrol presence to visually inspect for safety hazards, equipment anomalies, and regulatory compliance—elements that cannot be fully automated today. Current AI vision systems could assist with some inspections if cameras are pre-positioned, but the comprehensive patrolling duty demands human presence and judgment.
Task automatabilityclaude-sonnet-51/5This requires physical patrolling of tanks and units, visual/physical inspection, and on-site judgment that current AI cannot perform end-to-end without robotics and sensor infrastructure far beyond typical deployment.
Adoption barriersclaude-haiku-4-5-202510014/5Petroleum refinery operations are heavily regulated by EPA, OSHA, and state authorities; many regulations explicitly require human inspection and documentation. Liability for missed safety issues is severe, and many refineries operate in union environments with strong human-labor agreements. These regulatory and contractual barriers significantly impede full automation.
Adoption barriersclaude-sonnet-54/5Safety-critical operations in regulated environments (oil/gas) typically require certified personnel for compliance verification and liability reasons, creating strong regulatory and organizational barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Installing and maintaining distributed sensor networks and autonomous monitoring systems for refinery tank farms is capital-intensive and operationally complex. The all-in cost of such infrastructure and oversight often exceeds the salary of a patrol operator, especially for smaller or existing facilities without such systems.
Cost vs. human wageclaude-sonnet-52/5Sensor and monitoring systems can reduce some manual checks cheaply, but full patrol replacement would require costly sensor networks, robotics, and integration, keeping overall cost comparable to or higher than human labor for full task coverage.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed IoT and SCADA systems can monitor tank levels remotely, but they are supervisory aids rather than replacements for patrol operations. No fully autonomous system reliably performs the complete patrol function (visual inspection, hazard detection, regulatory compliance verification) in refinery environments, which remain primarily human-supervised.
Technical feasibility todayclaude-sonnet-52/5Some sensor-based tank monitoring and SCADA/IoT systems exist and are deployed, but full physical patrol and safety/compliance verification by AI is not a mature production capability.

Control or operate manifold and pumping systems to circulate liquids through a petroleum refinery.

20

CI 20–20 · exposure 25 · augmentation 50 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Refining is a laggard sector for autonomous AI adoption due to safety regulations, union presence, high cost of failures, and the capital-intensive nature of physical automation. Adoption has been limited to monitoring and alerting rather than autonomous control.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a capital-intensive, physically-oriented sector with slower digitization and AI adoption compared to information/professional services, though DCS and predictive maintenance tools are used.
Augmentation potentialclaude-haiku-4-5-202510013/5AI systems can assist operators by providing real-time analytics, anomaly detection, predictive maintenance alerts, and optimization recommendations, raising situational awareness and efficiency. However, the operator remains the decision-maker on manifold settings and pump sequences, with AI offering useful but not transformative assistance.
Augmentation potentialclaude-sonnet-53/5AI-based predictive analytics, anomaly detection, and control system optimization can meaningfully assist operators in monitoring and adjusting pump/manifold systems, improving efficiency and safety while humans remain in control.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could theoretically monitor sensor data and trigger predefined pump operations, the task requires real-time decision-making under variable conditions, emergency response, and safety-critical coordination across interconnected systems. Current AI systems lack the embodied control, equipment-specific domain adaptation, and fail-safe reasoning needed for reliable end-to-end operation at equal quality.
Task automatabilityclaude-sonnet-52/5This involves physical control of industrial equipment (manifolds, pumps) via control rooms and field valves; while DCS/SCADA automation exists, full end-to-end operation including physical checks and emergency response is not replaceable by off-the-shelf AI today.”,
Adoption barriersclaude-haiku-4-5-202510015/5Petroleum refining is heavily regulated by EPA, OSHA, and industry standards; operators must be licensed and certified. Liability exposure for equipment damage, product loss, or safety incidents is severe. Regulatory frameworks require human accountability and sign-off on critical operations, creating hard legal and compliance barriers to autonomous substitution.
Adoption barriersclaude-sonnet-55/5Refinery operations are heavily regulated (OSHA, EPA, process safety management) and require certified operators to be accountable for hazardous material handling, making full automation legally and organizationally very difficult.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration costs for AI control in a safety-critical refinery environment (validation, redundancy, regulatory compliance, fallback systems) substantially exceed the cost of a refinery operator's loaded salary, making autonomous AI more expensive than human operation.
Cost vs. human wageclaude-sonnet-52/5Existing process control automation reduces some labor costs, but comprehensive AI-driven autonomous operation would require expensive integration, redundant safety systems, and human oversight, keeping costs comparable to trained operators.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI systems reliably operate refinery manifold and pumping systems autonomously in production. Supervisory control systems exist but require human operators for critical decisions, fault recovery, and system optimization—AI is used only for monitoring and advisory roles, not autonomous operation.
Technical feasibility todayclaude-sonnet-52/5Automated control systems and PLCs handle routine circulation, but human operators remain essential for monitoring, judgment calls, and safety interventions in actual refineries; no AI product performs the full task autonomously.

Maintain and repair equipment, or report malfunctioning equipment to supervisors so that repairs can be scheduled.

18

CI 7–28 · exposure 13 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Refineries have deployed monitoring and sensor systems but adoption of autonomous repair remains minimal. Pilots for predictive maintenance exist, but the sector is cautious and regulatory-bound; human operator displacement in this domain is slow.
Sector adoption velocityclaude-sonnet-52/5Oil and gas/refining is a heavy industrial sector with historically slower digitization and AI adoption for physical maintenance tasks compared to information-sector work, though predictive maintenance pilots are growing.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-powered diagnostics and predictive maintenance alerts can assist operators in prioritizing repairs and understanding equipment trends, improving decision speed. However, the hands-on nature of repair work limits augmentation impact compared to cognitively intensive tasks.
Augmentation potentialclaude-sonnet-53/5AI-driven predictive maintenance and anomaly detection systems can flag equipment issues and prioritize repair scheduling, meaningfully assisting the reporting and diagnostic portion of this task even though physical repair remains manual.
Task automatabilityclaude-haiku-4-5-202510012/5Detection of malfunction and initial diagnostics could partially be automated via sensor monitoring, but the task requires physical repair work and hands-on judgment about equipment state that current AI systems cannot perform autonomously. The reporting component is automatable, but the core maintenance and repair functions remain heavily human-dependent.
Task automatabilityclaude-sonnet-51/5Physical inspection, hands-on maintenance and repair of industrial pump/refinery equipment requires manual dexterity and physical presence that current AI cannot replicate; only reporting communication could be marginally assisted.
Adoption barriersclaude-haiku-4-5-202510014/5Refinery operations are heavily regulated (EPA, OSHA, process safety) and require licensed or certified personnel to perform repairs and sign off on equipment status. Liability exposure for missed failures is high, and federal safety standards require human accountability for equipment maintenance in hazardous chemical environments.
Adoption barriersclaude-sonnet-54/5Safety regulations, certification requirements for refinery operations, and liability for equipment failure in hazardous environments create strong barriers to full automation of maintenance and repair decisions.
Cost vs. human wageclaude-haiku-4-5-202510012/5Monitoring systems reduce some labor costs, but the specialized technician wages for actual repair work, combined with integration and AI oversight costs, mean total automation remains comparable or more expensive than employing skilled human operators.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical repair labor, so there is no comparable AI cost basis; any AI use would be additive (monitoring) rather than replacing the human cost of hands-on work.
Technical feasibility todayclaude-haiku-4-5-202510012/5Sensor-based monitoring and automated alerting systems exist in industrial settings, but end-to-end repair capability and the judgment-heavy diagnosis of complex refinery equipment faults remain beyond current deployment. Monitoring products show material blind spots with equipment anomalies.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical equipment repair or maintenance in refineries; sensor-based predictive maintenance systems exist but do not perform the hands-on task itself.

Conduct general housekeeping of units, including wiping up oil spills and performing general cleaning duties.

10

CI 5–15 · exposure 0 · augmentation 13 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refinery operators are traditionally slower adopters of automation in front-line hazardous-duty roles due to regulatory conservatism, safety criticality, and the physical specificity of each facility's layout and equipment.
Sector adoption velocityclaude-sonnet-51/5Physical, low-digitization housekeeping tasks in industrial/refinery settings show minimal AI or robotic adoption to date.
Augmentation potentialclaude-haiku-4-5-202510012/5While monitoring systems could alert workers to spill locations or volume, AI offers limited augmentation for the hands-on physical cleaning task itself without substantial robotics integration already under heavy barriers.
Augmentation potentialclaude-sonnet-51/5AI offers essentially no meaningful assistance for physical spill cleanup and general cleaning duties performed manually by operators.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical manipulation of cleaning equipment in hazardous, unstructured refinery environments with variable spill locations and compositions. Current AI lacks embodied robotics capable of safely navigating and cleaning industrial facilities at scale.
Task automatabilityclaude-sonnet-51/5This is a physical manual labor task requiring cleaning oil spills and equipment maintenance in a hazardous industrial environment; no current AI can perform this physical work.
Adoption barriersclaude-haiku-4-5-202510014/5Refineries operate under strict EPA and OSHA regulations with documented safety protocols; oil spill cleanup and housekeeping in hazardous environments face liability, worker safety mandates, and legal requirements for human oversight and certification.
Adoption barriersclaude-sonnet-52/5No licensing barrier specifically prevents automation of cleaning, though safety regulations in hazardous refinery environments create some procedural friction for introducing robotics.
Cost vs. human wageclaude-haiku-4-5-202510011/5Specialized industrial robots capable of hazardous environment cleaning remain significantly more expensive than human labor when accounting for deployment, maintenance, and insurance costs in refinery contexts.
Cost vs. human wageclaude-sonnet-51/5AI cannot perform this physical task at all, so there is no viable AI cost comparison; any robotic solution would be far more expensive than human labor for this task.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed commercial systems reliably perform general housekeeping and oil spill cleanup in active refinery settings. This requires physical presence, dexterity, and real-time hazard assessment beyond current robotic deployment maturity.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical housekeeping/cleaning in refinery environments; this requires robotic manipulation which remains research-stage for such unstructured hazardous tasks.

Coordinate shutdowns and major projects.

8

CI 0–16 · exposure 8 · augmentation 50 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refinery operations are capital-intensive, highly regulated, and risk-averse; adoption of autonomous coordination systems is minimal. The industry favors proven human expertise and incremental tool improvements over replacing coordinators with AI, and no public production deployments of autonomous shutdown coordination are evident.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a capital-intensive, safety-regulated, physically-oriented sector with slower AI adoption compared to information/professional services sectors.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist with scheduling optimization, real-time alert synthesis, and documentation, helping human coordinators work more efficiently. However, augmentation is limited to support tasks; AI cannot meaningfully augment the core judgment and decision-making that shutdown coordination requires.
Augmentation potentialclaude-sonnet-53/5AI can assist with scheduling optimization, predictive maintenance data, and project timeline analysis, but the core coordination, negotiation, and safety oversight remain human-led.
Task automatabilityclaude-haiku-4-5-202510012/5Coordinating shutdowns and major projects requires real-time decision-making, stakeholder communication, and dynamic problem-solving across multiple interdependent systems. While AI could assist with scheduling and documentation, the core coordination—managing human teams, responding to unforeseen complications, and making judgment calls—remains fundamentally human-dependent and cannot achieve the ≥50% time-saving threshold end-to-end.
Task automatabilityclaude-sonnet-51/5Coordinating shutdowns and major projects requires physical site presence, real-time cross-team judgment, and safety-critical decision-making in a hazardous industrial environment that current AI cannot perform end-to-end.tered.
Adoption barriersclaude-haiku-4-5-202510015/5Refinery shutdowns are heavily regulated under OSHA, EPA, and process-safety management rules, and legal liability for errors is severe. A licensed, accountable human must ultimately authorize and oversee shutdown coordination, creating a hard regulatory and liability barrier to full automation.
Adoption barriersclaude-sonnet-54/5Shutdown coordination involves safety regulations, liability for catastrophic failure, and often requires certified operators and engineers to sign off on procedures, creating strong barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI systems capable of coordinating high-stakes refinery shutdowns—including safety monitoring, regulatory compliance, and liability oversight—exceeds the cost of experienced human coordinators who perform this work today.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this coordination role, so cost comparison favors the human by default; any AI tools used are supplementary, not replacements.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs end-to-end shutdown or major-project coordination in refinery operations at scale. The safety-critical nature, legal liability, and need for real-time human oversight mean this remains research-stage; existing systems cannot independently handle the coordination demands.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product autonomously coordinates refinery shutdowns or major capital projects today; this remains firmly in the domain of human project managers and operators.

Collect product samples by turning bleeder valves, or by lowering containers into tanks to obtain oil samples.

5

CI 5–5 · exposure 0 · augmentation 25 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refining is capital-intensive and heavily regulated; while some remote monitoring has been deployed, hands-on sampling tasks remain performed by human operators due to safety standards, regulatory compliance, and the complex, hazardous physical environment.
Sector adoption velocityclaude-sonnet-51/5Refining and heavy industrial operations are slow to adopt AI/robotics for physical hands-on tasks like manual sampling, lagging far behind digital-first sectors.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could provide some assistance through predictive analytics on when samples are needed or anomaly detection in sample composition, but the core task of physically collecting samples offers limited augmentation opportunity beyond basic scheduling or data logging.
Augmentation potentialclaude-sonnet-52/5AI could assist with scheduling, logging, or analyzing sample data post-collection, but offers minimal assistance to the physical act of sample collection itself.
Task automatabilityclaude-haiku-4-5-202510011/5The task requires physical manipulation of valves and containers in hazardous refinery environments where real-time sensory feedback and safety-critical decision-making are essential. Current AI systems cannot perform this hands-on, location-specific physical action end-to-end.
Task automatabilityclaude-sonnet-51/5This is a physical manipulation task requiring on-site presence to operate valves and lower containers into tanks; no AI system can perform the physical sampling itself.atile.
Adoption barriersclaude-haiku-4-5-202510014/5Refineries operate under strict EPA, OSHA, and Process Safety Management regulations that mandate trained, certified personnel to handle sampling and valve operations; liability and safety certification requirements create strong legal and operational barriers to full automation.
Adoption barriersclaude-sonnet-54/5Handling hazardous petroleum products involves safety, environmental, and regulatory requirements often necessitating trained, certified personnel physically present at the site.
Cost vs. human wageclaude-haiku-4-5-202510011/5Robotics capable of performing this task in a refinery setting would be extremely expensive to deploy, maintain, and oversee, far exceeding the cost of a trained operator performing the routine sampling.
Cost vs. human wageclaude-sonnet-51/5AI has no direct role in performing this physical task, so any AI-based approach would require costly robotic hardware exceeding current human labor costs for this narrow task.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product can autonomously collect oil samples by physically operating bleeder valves or lowering containers into tanks in production refineries. This remains a human-performed task with no viable commercial automation solution.
Technical feasibility todayclaude-sonnet-51/5No deployed AI/robotic product performs this specific physical sampling task in refineries today; automated sampling systems exist but are hardware/engineering solutions, not AI-driven task substitutes.

Clamp seals around valves to secure tanks.

5

CI 5–5 · exposure 0 · augmentation 25 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refineries are capital-intensive but technologically conservative in safety-critical operations; adoption of autonomous physical manipulation systems remains negligible in the sector. Existing automation focuses on monitoring and control systems, not manual sealing tasks.
Sector adoption velocityclaude-sonnet-51/5Refinery and industrial physical operations sectors show very low AI/robotic adoption for manual mechanical tasks like this, remaining highly manual and slow to digitize.
Augmentation potentialclaude-haiku-4-5-202510012/5Limited augmentation potential exists; AI could provide real-time guidance on seal integrity via computer vision or alert operators to valve anomalies, but the physical act of clamping itself offers little room for AI assistance without full automation. Remote monitoring aids oversight but does not meaningfully raise the operator's productivity on the core task.
Augmentation potentialclaude-sonnet-52/5AI could support with checklists, sensor monitoring, or predictive alerts for seal integrity, but offers minimal direct assistance to the physical clamping action itself.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical manipulation in a hazardous industrial environment—clamping seals around valves demands dexterity, force calibration, and tactile feedback that current AI/robotic systems cannot reliably perform end-to-end in unstructured refinery settings. No commercial AI system performs this task today.
Task automatabilityclaude-sonnet-51/5This is a physical manipulation task requiring hands-on placement and tightening of clamp seals on valves in a hazardous industrial environment; no current AI system can perform this physical action end-to-end.'
Adoption barriersclaude-haiku-4-5-202510014/5This task operates in a safety-critical, hazardous environment subject to EPA, OSHA, and refinery-specific regulations; liability for improper sealing (leaks, spills) falls on the operator and facility, creating strong legal and safety barriers to automation or delegation to unproven systems. Human verification and sign-off are typically required.
Adoption barriersclaude-sonnet-54/5Safety and regulatory requirements in refinery operations mandate trained, often certified personnel to handle tank sealing and valve security due to high liability and hazardous material risks.
Cost vs. human wageclaude-haiku-4-5-202510011/5Any capable industrial robotic system (if one existed for this task) would cost hundreds of thousands of dollars plus integration; a refinery operator performing this task costs far less per unit output, making AI/robotics prohibitively expensive relative to human labor.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute for the physical act, so any AI-based approach (e.g., robotics) would require expensive specialized hardware far exceeding human labor costs for this simple manual task.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs valve seal clamping in production refinery environments. This is fundamentally a physical manipulation task requiring industrial robotics, which remains limited to highly structured, repetitive scenarios—not the variable geometry and safety-critical nature of valve sealing.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs physical valve sealing; this remains a manual task requiring human dexterity and presence at the equipment.

Clean interiors of processing units by circulating chemicals and solvents within units.

4

CI 0–7 · exposure 0 · augmentation 25 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Refinery operations remain highly conservative and heavily regulated sectors with low digitization of core process tasks. Automation of chemical circulation cleaning is not a focus of industry investment relative to other opportunities.
Sector adoption velocityclaude-sonnet-52/5Oil and gas refining is a heavy industrial sector with slower digitization and physical-task automation adoption compared to information/professional services sectors, though some process control automation exists.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could potentially assist by monitoring chemical circulation parameters or predicting optimal cleaning cycles, but the core physical task of circulating chemicals requires human operators in the loop for safety and regulatory compliance.
Augmentation potentialclaude-sonnet-52/5AI/software can assist with monitoring sensor data, scheduling maintenance windows, and flagging anomalies during cleaning cycles, but it does not materially transform the hands-on physical execution of this task.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical circulation of chemicals within processing units, involving hands-on equipment operation, chemical handling, and real-time sensory feedback. Current AI cannot physically manipulate chemical systems or operate pumps and valves in the real world without human intervention.
Task automatabilityclaude-sonnet-51/5This is a physical cleaning task requiring circulating chemicals/solvents through industrial equipment, which requires physical presence, valve operation, and hands-on monitoring that no AI system can perform end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5Petroleum refining operates under strict OSHA, EPA, and process safety regulations requiring licensed operators to directly oversee and perform hazardous chemical operations. Human presence and authorization are legally mandated for chemical handling in refineries.
Adoption barriersclaude-sonnet-54/5Refinery cleaning operations involve hazardous chemicals, strict safety/environmental regulations, and require trained, often certified operators physically present, creating strong procedural and regulatory barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of deploying robotic systems capable of safely handling chemicals in refinery environments, plus integration and maintenance, far exceeds the cost of human operators performing this task. Specialized industrial robotics for hazardous chemical handling remain prohibitively expensive.
Cost vs. human wageclaude-sonnet-51/5AI cannot perform the physical circulation and hands-on procedural steps, so there is no viable AI substitute cost to compare; human labor plus existing automated control systems remain necessary.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can autonomously perform the physical circulation of chemicals and solvents within refinery processing units. This requires embodied robotics in hazardous industrial environments, which is not a mature production capability.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical chemical circulation cleaning of refinery processing units; this remains a manual/physical operational task performed by human operators with control-system support.

Inspect pipelines, tightening connections and lubricating valves as necessary.

3

CI 0–5 · exposure 0 · augmentation 38 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5The refinery and pipeline sectors remain capital-intensive, safety-critical environments where human operators are legally required and adoption of fully autonomous maintenance automation is extremely slow. Most automation to date focuses on remote monitoring rather than autonomous physical maintenance.
Sector adoption velocityclaude-sonnet-51/5Oil and gas refinery operations are a physical, heavy-industry sector with historically slow digitization and robotic automation adoption for manual maintenance tasks.
Augmentation potentialclaude-haiku-4-5-202510012/5Digital inspection tools and remote monitoring systems can assist operators in locating problem areas, but AI does not meaningfully augment the core physical tasks of tightening connections or lubricating valves, which remain manual and operator-dependent.
Augmentation potentialclaude-sonnet-53/5AI-enabled sensors, predictive maintenance analytics, and inspection drones/robots can help prioritize and flag which pipelines need attention, augmenting the operator's decision-making even though the physical task remains manual.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical manipulation of valves, connections, and equipment in a potentially hazardous environment. Current AI systems cannot independently perform the tactile work of tightening connections or applying lubricants, nor can they reliably navigate and inspect complex industrial pipelines without human oversight.
Task automatabilityclaude-sonnet-51/5This requires physical manipulation of valves and connections in a hazardous industrial environment; no off-the-shelf AI or robotic system can perform this end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5Petroleum operations are heavily regulated (EPA, OSHA) with legal requirements for licensed operators to conduct and sign off on safety-critical maintenance. Environmental and workplace safety liability is asymmetric, making substitution of human judgment with automation legally and organizationally difficult.
Adoption barriersclaude-sonnet-54/5Safety regulations, hazardous environment protocols, and liability for pipeline integrity failures create strong barriers requiring trained, often certified personnel to perform hands-on inspection and maintenance.
Cost vs. human wageclaude-haiku-4-5-202510011/5Specialized robotic systems capable of performing these physical maintenance tasks would have high capital and operational costs far exceeding the loaded wage of a skilled pump operator performing routine inspections.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing the physical labor, so cost comparison favors the human worker who must be employed regardless.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI products reliably perform independent pipeline inspection with tightening and lubrication in live refinery settings. While remote inspection drones exist, they lack the dexterity and autonomous decision-making to execute the full maintenance task without human direction.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously inspects pipelines and physically tightens connections or lubricates valves; sensor-based monitoring exists but the physical maintenance action is not automated.

Related occupations — Production

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.