First-Line Supervisors of Security Workers

33-1091.00

Median wage $55,940/yr81,480 employed (US)Rank #366 of 923 scored · top 40% by substitution

Directly supervise and coordinate activities of security workers and security guards.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution31

Exposure29

Augmentation58

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

21 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

10%

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%28

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

Technical feasibility todayw 20%31

panel mean rating 2.2/5 → substitution pressure 31/100

Cost vs. human wagew 15%32

panel mean rating 2.3/5 → substitution pressure 32/100

Adoption barriersw 20%inverted — strong barriers lower the score40

panel mean rating 3.4/5 (barrier strength) → substitution pressure 40/100

Sector adoption velocityw 10%26

panel mean rating 2.0/5 → substitution pressure 26/100

Task breakdown (21 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.

Log items distributed to persons, such as keys and key cards.

CI 81–89 · exposure 84 · augmentation 75 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Security and facility management sectors have widely adopted automated access control and inventory tracking systems over the past decade. Most medium to large organizations already employ digital logging for key and badge distribution rather than manual processes.
Sector adoption velocity	claude-sonnet-5	3/5	Security services and facility management are moderate adopters of digital access systems; many smaller operations still use manual logs, keeping overall adoption uneven.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered logging systems augment supervisory work by providing real-time dashboards, anomaly alerts, and audit trail generation, enabling supervisors to focus on exception handling and security analysis rather than manual data entry and record-keeping.
Augmentation potential	claude-sonnet-5	4/5	AI-enabled access control and digital logging systems substantially reduce manual record-keeping burden and error while supervisors retain oversight of exceptions and audits.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI systems can reliably handle logging, categorization, and record-keeping for physical items and access credentials with minimal human intervention. The task involves straightforward data entry and tracking, which OCR and database management systems automate efficiently, achieving >50% time savings.
Task automatability	claude-sonnet-5	4/5	Logging distributed items like keys and key cards is a simple structured data-entry task that can be handled by digital access control systems or basic software with minimal human involvement.dollarThe core logging action is easily automatable, though issuance itself may still involve a human handing over the item.
Adoption barriers	claude-haiku-4-5-20251001	2/5	While some organizations may prefer human oversight for security-sensitive processes, there are no legal licensing requirements preventing automated logging systems. Regulatory compliance can typically be achieved through audit trails and reporting features in the automation itself.
Adoption barriers	claude-sonnet-5	1/5	No licensing, liability, or regulatory requirement mandates a human specifically perform this administrative logging task.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated logging systems have negligible per-transaction costs compared to the loaded hourly wage of a first-line supervisor manually documenting each key and card distribution. Implementation is one-time; operational cost is orders of magnitude lower than human labor.
Cost vs. human wage	claude-sonnet-5	5/5	Automated logging via badge/key-card systems costs a small fraction of a supervisor's time spent on manual record-keeping, especially at scale.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Deployed access control systems, inventory management software, and RFID/barcode logging solutions already perform this task in production across organizations worldwide. These systems reliably track and log key distribution at scale with minimal error rates.
Technical feasibility today	claude-sonnet-5	5/5	Electronic key/card management systems and access control software already log issuance, returns, and timestamps reliably in production across many facilities today.

Order materials or supplies, such as keys, uniforms, and badges.

CI 72–84 · exposure 75 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Procurement automation is widely adopted in large enterprises and government agencies already using e-procurement platforms and RPA. Smaller security firms lag, but the trend is clearly toward automated ordering systems.
Sector adoption velocity	claude-sonnet-5	3/5	Security services firms are often smaller or less digitized than finance or tech, so while procurement automation exists broadly, adoption within this specific sector is moderate rather than cutting-edge.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by suggesting reorder quantities based on inventory trends, flagging stock-outs, and auto-drafting orders for supervisor review, meaningfully speeding up the ordering process while the human retains decision authority.
Augmentation potential	claude-sonnet-5	4/5	AI-assisted inventory and procurement tools can significantly streamline reorder decisions, demand forecasting, and vendor communication for supervisors who remain responsible for final decisions.
Task automatability	claude-haiku-4-5-20251001	4/5	Ordering materials is highly routine and structured, involving inventory checks, vendor selection, and purchase order generation—all tasks AI systems can perform reliably. With integration into existing supply chain systems, AI could handle 80–90% of this task end-to-end, meeting the 50% time-saving threshold.
Task automatability	claude-sonnet-5	4/5	Ordering standard supplies like keys, uniforms, and badges is a routine procurement task that can largely be automated via inventory management systems and reorder rules with a human approving purchases.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Few barriers exist: no licensing requirement, no regulatory mandate for human authorization, and minimal liability risk for standard supply orders. Some organizations may require supervisory approval workflows, adding modest friction but not preventing automation.
Adoption barriers	claude-sonnet-5	1/5	There is no licensing, regulatory, or human-contact requirement tied to ordering uniforms or badges; it's a routine administrative function.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI-driven procurement is multiple orders of magnitude cheaper per transaction than human clerk time, especially at scale. Integration costs amortize quickly over hundreds or thousands of orders.
Cost vs. human wage	claude-sonnet-5	4/5	Automated procurement systems cost far less per transaction than a supervisor's time spent manually tracking stock and placing orders, though some setup and vendor integration costs exist.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature e-procurement and inventory management systems with AI-assisted ordering are deployed in production across many organizations. API integration with vendor systems and internal inventory databases is standard, though some manual oversight of quantities and specifications typically remains.
Technical feasibility today	claude-sonnet-5	4/5	Procurement and inventory management software with automated reordering, vendor integration, and purchase order generation is widely deployed in production across industries today.

Monitor closed-circuit television cameras.

CI 46–72 · exposure 62 · augmentation 88 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Large enterprises and critical infrastructure operators are actively piloting and deploying AI CCTV monitoring, but adoption remains uneven. Small and mid-sized security operations lag significantly, keeping overall velocity in the moderate range.
Sector adoption velocity	claude-sonnet-5	3/5	Security and surveillance technology adoption is growing steadily but is uneven, more common in large enterprises and cities than small businesses, so overall diffusion is moderate.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly enhances human productivity by automating alert triage, highlighting suspicious activity, and enabling one supervisor to monitor more feeds. The human retains judgment over response, making this a strong augmentation scenario.
Augmentation potential	claude-sonnet-5	5/5	AI significantly boosts human monitoring capacity by pre-filtering events, reducing fatigue, and enabling one supervisor to effectively oversee many more feeds.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can detect anomalies, motion, and known threats in CCTV feeds with reasonable accuracy, automating routine monitoring and alerting. However, the task typically requires judgment about context, false positives, and integration with broader security decisions, limiting end-to-end automation to roughly 50% time savings.
Task automatability	claude-sonnet-5	4/5	Modern video analytics with AI-based motion/object/anomaly detection can automate a large share of continuous monitoring, flagging events for human review rather than requiring constant human watching.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Regulatory requirements and liability concerns create moderate friction; security incidents traced to automation failures expose operators to legal and reputational risk. However, no license is strictly required to deploy AI monitoring, and many organizations are piloting such systems.
Adoption barriers	claude-sonnet-5	2/5	Some liability concerns exist for security decisions and escalation, but automated CCTV monitoring itself is largely unregulated and already common as a supplementary tool.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	High-quality CCTV AI systems require significant infrastructure, integration, and ongoing oversight costs. While inference is cheap, the all-in cost (hardware, licensing, human validation) remains comparable to or higher than a security worker's loaded wage for many deployments.
Cost vs. human wage	claude-sonnet-5	4/5	AI-based camera monitoring software costs a fraction of paying a guard to watch screens continuously, especially at scale across many camera feeds.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed video analytics and AI monitoring systems exist in production (e.g., anomaly detection, person tracking), but they generate material false positive rates and require human oversight. Most real-world deployments serve as assistive tools rather than fully autonomous replacements.
Technical feasibility today	claude-sonnet-5	4/5	Commercial video analytics platforms (e.g., Verkada, Avigilon, Genetec) are widely deployed in production for intrusion detection, loitering, and object recognition, though they still require human verification for ambiguous events.

Write reports documenting observations made while on patrol.

CI 25–60 · exposure 45 · augmentation 88 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security firms operate across private and public sectors with varied digitization; most still rely on human-written reports. While some firms experiment with mobile forms and templates, autonomous report generation remains rare in production security operations.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a lower-digitization, physically-oriented sector with slower AI adoption compared to information/finance sectors, though report drafting tools are slowly entering the market.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist by auto-completing templates, suggesting incident classifications, organizing observations, and drafting summaries that supervisors then refine—substantially accelerating report completion while the human retains full control and accountability.
Augmentation potential	claude-sonnet-5	5/5	AI is highly effective at drafting, organizing, and polishing reports from raw observational notes, significantly speeding up the writing process while the supervisor verifies content.
Task automatability	claude-haiku-4-5-20251001	2/5	Writing patrol reports requires subjective judgment about incident severity, context, and relevance that current AI cannot reliably automate. While AI can assist with formatting and minor data entry, security supervisors must interpret observations and decide what to document—tasks requiring human discretion and accountability.
Task automatability	claude-sonnet-5	4/5	Turning observational notes into structured incident/patrol reports is a well-bounded language task that current LLMs handle well, especially given dictated or bulleted input, though initial observation and judgment about what to include remains human.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security reports are legal documents used in investigations, court proceedings, and liability disputes; they must be signed by the reporting officer and reflect their observations. Regulatory and legal frameworks require human authorship and accountability, creating a strong barrier to full automation.
Adoption barriers	claude-sonnet-5	3/5	Reports may serve as legal/evidentiary documents requiring accuracy and accountability, so a human must review and sign off, creating moderate liability-driven barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of AI systems (inference, integration, error-checking) is not dramatically cheaper than having a supervisor spend 15–30 minutes writing a report, especially when human oversight is required to verify accuracy and completeness for legal and liability reasons.
Cost vs. human wage	claude-sonnet-5	4/5	Generating a written report from notes via AI costs a fraction of a cent to a few cents versus the supervisor's time, making it dramatically cheaper once inputs are captured.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No production systems currently generate security patrol reports autonomously at the quality required for legal/liability purposes. While LLMs can draft text, they cannot reliably assess what observations matter in a security context or ensure accuracy without substantial human review, making them unsuitable for independent deployment.
Technical feasibility today	claude-sonnet-5	3/5	Report-writing assistants and dictation-to-report tools exist in security/law enforcement software, but widespread reliable deployment specifically for patrol supervisor reports is still limited and often manual.

Write and present department budgets to upper management or other stakeholders.

CI 25–52 · exposure 38 · augmentation 63 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security sector adoption of AI is slower than finance or tech; budget presentation is a periodic, high-stakes task that organizations approach cautiously. Pilots exist but production-level autonomous budget writing and presentation remains rare in security operations.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a relatively low-digitization, operations-heavy sector where AI adoption for administrative/managerial tasks lags behind finance or tech sectors.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by drafting budget templates, aggregating expense data, and suggesting narrative framing, which helps supervisors work faster. However, the human must still validate numbers, make strategic choices, and deliver the presentation, so augmentation is partial rather than transformative.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially help draft budget narratives, format data, model scenarios, and prepare talking points, meaningfully boosting productivity while the supervisor retains ownership of the presentation.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can draft budget narratives and compile financial data, writing and presenting budgets to upper management requires nuanced judgment about organizational priorities, stakeholder concerns, and strategic justification that goes beyond template generation. Current systems struggle with the full end-to-end context and persuasive communication needed for live executive presentations.
Task automatability	claude-sonnet-5	3/5	AI can draft budget documents, compile historical data, and generate narratives, but the actual presentation to stakeholders and defense of judgment calls requires human involvement, so only part of the task meets the time-saving bar.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Budget authority and sign-off typically rest with the human supervisor, who is personally accountable to upper management for budget assumptions and performance. This fiduciary and accountability requirement creates a strong legal and organizational barrier to full automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement forces a human to write or deliver a budget, but organizational norms expect the accountable supervisor to personally own and defend financial decisions to leadership.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	While AI-assisted drafting reduces some clerical work, the requirement for human supervisors to review, contextualize, and present budgets means total cost savings remain modest compared to a supervisor's loaded wage for this full task.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cut drafting time meaningfully but a supervisor still must review, tailor, and present the budget, so overall cost savings are moderate rather than order-of-magnitude given the labor is a small part of a manager's overall role.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools can assist with budget document generation and formatting, but no deployed product reliably handles the complete task of writing context-appropriate, strategically sound department budgets and delivering convincing presentations to executive audiences without substantial human oversight and revision.
Technical feasibility today	claude-sonnet-5	3/5	Products like spreadsheet AI assistants and generative writing tools are deployed for budget drafting and financial narrative generation, but they don't reliably handle organization-specific context or live presentation without heavy human editing.

Develop and document security procedures, policies, or standards.

CI 25–43 · exposure 33 · augmentation 75 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security management is moderately digitized but conservative; adoption of AI for policy generation remains limited to pilots and assistive use in forward-looking firms. Most security teams still rely on human supervisors and consultants for procedure development.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a physically-oriented, lower-digitization sector where AI adoption for documentation tasks is emerging but not yet deep or fast compared to information/professional services sectors.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist supervisors by generating initial drafts, flagging common policy gaps, suggesting compliant language, and organizing procedures—substantially accelerating the authoring process while the supervisor retains full control and accountability for final content.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully speed up drafting, formatting, and researching best-practice language for security policies, letting the supervisor focus on customization and approval, a clear productivity boost while human judgment remains central.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can draft security procedures and suggest policy language, but the task requires domain expertise, legal compliance knowledge, and organizational context that current systems handle inconsistently. A human must substantially review, customize, and validate any AI-generated output, limiting time savings below 50%.
Task automatability	claude-sonnet-5	3/5	AI can draft initial security policy documents and procedures based on templates and inputs, but requires substantial human review, site-specific knowledge, and judgment to finalize, so only partial time savings are realized end-to-end.6
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security policies often require sign-off by legal, compliance, and senior management; many organizations have regulatory obligations (HIPAA, SOC 2, etc.) mandating human accountability for documented procedures. Liability and error-cost asymmetry create strong friction against full automation.
Adoption barriers	claude-sonnet-5	3/5	There's no strict licensing requirement to write these documents, but liability, regulatory compliance obligations, and organizational sign-off requirements create moderate friction against fully automating final policy authorship.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI inference and integration costs for policy generation are low, but the human oversight and expert validation required to ensure legal compliance, liability protection, and organizational fit mean total cost remains comparable to having a supervisor draft policies directly.
Cost vs. human wage	claude-sonnet-5	3/5	AI drafting tools are cheap per document, but the need for expert review, site visits, and compliance checks by a supervisor keeps overall cost roughly comparable to a human doing the full task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While LLMs can generate policy templates and procedure outlines, no mature deployed product reliably produces organization-ready security documentation without expert human revision. Existing tools are primarily assistive drafting aids, not autonomous producers of compliance-grade policies.
Technical feasibility today	claude-sonnet-5	2/5	Generic document drafting tools exist and are used for policy writing, but no deployed product reliably produces complete, compliant, site-specific security procedures without heavy human editing and validation.

Schedule training or drills for emergencies, such as fires, bombs, and other threats.

CI 30–34 · exposure 25 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security operations remain relatively conservative sectors with slower digital transformation. While some large enterprises use scheduling software, adoption of AI-driven autonomous scheduling for emergency drills is still in pilot phases rather than widespread production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a relatively low-digitization, physically-oriented sector with slower AI adoption for operational planning tasks compared to information-sector functions.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist supervisors by proposing optimal drill schedules, suggesting compliance deadlines, and tracking completion across teams. However, human supervisors must retain decision authority on emergency preparedness strategy and threat assessment, limiting full transformation potential.
Augmentation potential	claude-sonnet-5	3/5	AI can assist by suggesting drill schedules, generating training materials, and tracking compliance calendars, providing useful support while the supervisor retains final planning responsibility.
Task automatability	claude-haiku-4-5-20251001	2/5	Scheduling training or drills is fundamentally an administrative task involving calendar coordination and logistical planning. While AI can assist with scheduling optimization and draft coordination, the task requires human judgment about threat assessment priorities, staff readiness, budget constraints, and organizational context that current AI systems cannot reliably capture end-to-end.
Task automatability	claude-sonnet-5	2/5	AI can help draft schedules and training content, but the task involves coordinating with staff, physical logistics, and organizational judgment that requires human oversight and cannot be fully automated end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Emergency preparedness decisions carry liability implications and organizational accountability. Supervisors are typically required to personally oversee and authorize training schedules due to safety and legal responsibility, creating moderate friction against full automation.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement for scheduling itself, but organizational policies, safety compliance obligations, and human accountability for emergency preparedness create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI-assisted scheduling tools are comparable in cost to a supervisor's time spent scheduling, but the supervisor must still review, validate, and oversee the output, limiting cost savings to roughly equal trade-offs rather than significant savings.
Cost vs. human wage	claude-sonnet-5	2/5	While AI scheduling assistance is cheap, the overall task still requires human coordination, facility knowledge, and compliance judgment, limiting cost savings versus a supervisor's time.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature production systems reliably perform independent scheduling of emergency drills at scale. While calendar and project management tools exist and AI can draft schedules, they require substantial human oversight for security-critical decisions about emergency preparedness timing and content.
Technical feasibility today	claude-sonnet-5	2/5	Scheduling and calendar tools with AI assistance exist, but no deployed product autonomously plans and executes emergency drill scheduling reliably in production for security operations.

Recruit, interview, and hire security personnel.

CI 25–37 · exposure 30 · augmentation 50 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Mid-sized adoption in corporate HR and security firms, with resume screening and scheduling tools in production, but interview and final decision-making remain human-driven. Security industry has been slower to adopt advanced automation compared to tech and finance sectors.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a lower-digitization, physical-labor-adjacent sector where HR AI adoption for hiring lags behind knowledge-work sectors despite general HR-tech uptake.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by filtering candidates, aggregating résumé data, and flagging red flags, reducing supervisor time on administrative steps. However, augmentation is limited to candidate pre-screening; human judgment on interview quality and hire fit remains essential and difficult to enhance with AI.
Augmentation potential	claude-sonnet-5	3/5	AI tools can meaningfully assist with resume screening, job description drafting, and scheduling, improving supervisor efficiency while humans retain interview and final decision responsibilities.
Task automatability	claude-haiku-4-5-20251001	2/5	Recruitment posting and initial résumé screening can be partially automated, but the core tasks—conducting substantive interviews, assessing soft skills, cultural fit evaluation, and final hiring decisions—require human judgment. Current AI can assist with candidate filtering but cannot reliably end-to-end replace the full hiring process with 50% time savings at equal quality.
Task automatability	claude-sonnet-5	2/5	AI can help draft job postings and screen resumes, but interviewing and final hiring decisions require human judgment, cultural fit assessment, and legal accountability that current systems cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Legal and liability barriers are substantial: employment law compliance, discrimination risk, potential security vetting requirements (background checks, clearances), and organizational accountability for hiring decisions make it difficult to fully delegate to AI without human sign-off. Supervisors face personal and organizational liability for poor hires.
Adoption barriers	claude-sonnet-5	3/5	Hiring decisions carry legal/liability exposure (discrimination law, background checks, licensing verification for security personnel) and typically require human sign-off, creating moderate but not absolute barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI recruitment tools reduce cost on screening and scheduling components, but human supervisors must still conduct interviews and make final decisions, limiting cost advantage. The all-in cost (tool subscription, oversight, error correction) is comparable to or exceeds the cost of a supervisor spending several hours on hiring.
Cost vs. human wage	claude-sonnet-5	2/5	Some cost savings exist from AI-assisted screening, but human interviewers and decision-makers remain necessary, keeping overall cost comparable to fully human-run hiring processes.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	AI tools (resume screening, interview scheduling, skill-matching) are deployed in HR platforms today, but with material limitations: bias concerns, poor evaluation of communication and interpersonal fit for security roles, and reliance on human oversight to avoid hiring errors. No production system fully automates security personnel hiring.
Technical feasibility today	claude-sonnet-5	2/5	ATS and AI screening tools are deployed widely for resume filtering, but reliable end-to-end AI-driven interviewing and hiring decisions for security roles are not standard production practice.

Monitor and authorize entry of employees, visitors, or other persons.

CI 25–35 · exposure 30 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While large enterprises (finance, tech, corporate campuses) have deployed AI-assisted entry systems, human security workers remain the legal and operational decision-point in most settings. Adoption is concentrated in high-security and well-resourced sectors; most organizations rely on human guards making the final authorization call.
Sector adoption velocity	claude-sonnet-5	2/5	Physical security and facilities management sectors adopt digitization slowly compared to information/professional services; automated access systems are common but full replacement of supervisory roles is rare and pilots for AI-driven security decision-making are nascent.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered identity verification and automated alerts can substantially assist supervisors by pre-screening visitors, flagging anomalies, and maintaining audit logs, freeing supervisors to focus on edge cases and judgment calls. This tool-layer assistance genuinely raises supervisor productivity without requiring full automation of the authorization decision.
Augmentation potential	claude-sonnet-5	4/5	AI-powered surveillance analytics, anomaly detection, and automated visitor logging significantly help supervisors monitor more entry points and flag issues faster, meaningfully boosting productivity while the human retains authorization authority.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with identity verification (facial recognition, badge scanning), the task requires contextual judgment about who should be authorized, exceptions, and real-time decision-making that involves discretion and accountability. End-to-end automation would require eliminating human oversight, which raises liability and security concerns that prevent the ≥50% time-saving bar from being consistently met.
Task automatability	claude-sonnet-5	2/5	Physical entry monitoring and authorization can be partially handled by badge systems, cameras, and access control software, but the supervisory judgment role (assessing suspicious behavior, exceptions, escalation) still requires human presence and decision-making.You cannot fully replace the human supervisor end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Physical security, legal liability for breaches, and potential regulatory requirements (depending on facility type and jurisdiction) create strong adoption barriers. Organizations face reputational and legal risk if an AI system grants unauthorized access or denies legitimate entry, creating pressure to retain human sign-off and accountability.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement specifically for entry monitoring, but liability concerns (security breaches, safety), organizational policy, and need for human judgment in ambiguous situations create meaningful friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI infrastructure (cameras, recognition software, integration, maintenance, oversight personnel) remains expensive relative to a security worker's wage, especially when factoring in the liability and error-correction costs of incorrect denials or unauthorized access. The system as a whole is not yet cheaper than human supervision.
Cost vs. human wage	claude-sonnet-5	2/5	Access control hardware/software has upfront and maintenance costs comparable to or sometimes exceeding a supervisor's marginal wage contribution for this sub-task, and human oversight is still required for exceptions, keeping cost savings modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Facial recognition and access-control systems exist in production, but they function as tools requiring human authorization rather than independent decision-makers. No deployed AI system reliably makes final entry authorization decisions without human oversight; vendor systems still operate at the support layer, not the authoritative gate-keeper layer.
Technical feasibility today	claude-sonnet-5	3/5	Automated access control systems, facial recognition, and visitor management software are widely deployed and reliable for routine authorization, but exception handling and supervisory oversight of security personnel remain human-performed in production settings.

Monitor the behavior of security employees to ensure adherence to quality standards, deadlines, or procedures.

CI 25–30 · exposure 25 · augmentation 50 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security firms are adopting monitoring tools (cameras, time systems) but not deploying autonomous AI to replace supervisory judgment at scale. The sector remains cautious due to liability and the need for human accountability in hiring, discipline, and performance management.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a lower-digitization sector where AI adoption for direct personnel supervision remains nascent, largely limited to surveillance analytics rather than full supervisory automation.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-powered dashboards and anomaly detection (e.g., scheduling conflicts, attendance patterns, GPS deviations) can assist supervisors in prioritizing what to review. However, the core task of interpreting behavior and ensuring adherence still rests with the human supervisor, limiting transformative impact.
Augmentation potential	claude-sonnet-5	3/5	AI-powered video analytics, scheduling tools, and incident-tracking systems can help supervisors monitor compliance and flag issues, improving efficiency while the supervisor remains responsible for judgment and action.
Task automatability	claude-haiku-4-5-20251001	2/5	Monitoring behavior against standards requires subjective judgment about adherence, context-dependent decision-making, and real-time observation of complex human interactions. While AI can flag outliers in structured data (e.g., patrol logs, response times), the interpretation of behavioral nuance and quality standards typically requires human supervisory judgment that current systems cannot reliably automate end-to-end.
Task automatability	claude-sonnet-5	2/5	Behavioral supervision requires in-person observation, judgment about conduct, and contextual coaching that current AI cannot fully replicate end-to-end, though some monitoring aspects (logs, timestamps, camera review) could be partially assisted.detected.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security supervision often involves legal accountability for employee conduct, potential liability for missed violations, and in many jurisdictions regulatory or contractual requirements that a qualified human supervisor must review and sign off on personnel monitoring decisions.
Adoption barriers	claude-sonnet-5	3/5	While not licensed per se, supervisory roles often carry accountability, HR, and labor-relations requirements that necessitate human judgment and authority over employees, creating moderate organizational and legal friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current monitoring tools (video, time-tracking, access logs) require significant integration, human oversight to interpret results, and ongoing tuning. Combined with the human supervisor still needed for judgment, total cost remains comparable to or higher than direct human supervision.
Cost vs. human wage	claude-sonnet-5	2/5	AI monitoring tools add cost on top of still-needed human supervisors for judgment calls and personnel management, so total cost is not clearly lower than a human supervisor performing this task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some narrow aspects exist (e.g., video analytics for presence detection, time-tracking software), but deployed systems do not reliably assess behavioral adherence to procedures or quality standards at the supervisory level. Products are limited to data collection rather than the judgment component of monitoring.
Technical feasibility today	claude-sonnet-5	2/5	Video analytics and workforce management tools exist for flagging anomalies or compliance issues, but no deployed product autonomously supervises security staff behavior reliably at scale.

Advise employees in handling problems or resolving complaints from customers, tenants, detainees, or other persons.

CI 25–30 · exposure 25 · augmentation 50 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security and facility management sectors have historically lagged in AI adoption, and supervisory complaint-handling remains a high-touch, relationship-dependent function where organizations are cautious about removing human judgment. Adoption has been limited to tools rather than replacement systems.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a relatively low-digitization, physically-oriented sector where AI adoption for supervisory/interpersonal tasks remains nascent compared to information or finance sectors.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist supervisors by drafting response templates, flagging policy issues, or suggesting precedents from similar complaints, meaningfully improving their efficiency and consistency. However, the human supervisor must retain judgment on tone, fairness, and relationship dynamics.
Augmentation potential	claude-sonnet-5	3/5	AI can help supervisors by suggesting de-escalation techniques, drafting complaint responses, or summarizing policy guidance, meaningfully aiding but not replacing the human advisory interaction.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help draft responses to complaints and suggest handling procedures, the task fundamentally requires understanding nuanced interpersonal dynamics, emotional context, and judgment about fairness that current systems struggle with reliably. Most of the time-critical value comes from human-to-human conflict resolution and credibility, which AI cannot fully replace.
Task automatability	claude-sonnet-5	2/5	This requires situational judgment, real-time interpersonal coaching, and knowledge of specific site policies and personalities, which current AI cannot fully replicate end-to-end despite being able to draft generic guidance.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Supervisory roles carry legal and organizational accountability for complaint resolution, employee relations, and adherence to labor/detention standards. Liability falls on the human supervisor, and most organizations require documented human sign-off on personnel and complaint matters, creating strong barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement mandates a human specifically for this advisory task, but liability concerns, security protocols, and the need for trusted on-site authority create meaningful organizational friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI assistance for generating complaint-handling templates or advice is inexpensive, but it still requires human supervisory review, decision-making, and follow-up, making the all-in cost comparable to direct human performance rather than substantially cheaper.
Cost vs. human wage	claude-sonnet-5	2/5	A human supervisor's advisory role is embedded in existing labor cost with no separate line item, while AI tools would add licensing/integration costs without replacing the supervisory role itself.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably handles this end-to-end in security supervisory contexts. Chatbots and LLMs can suggest responses, but they lack context about specific relationships, organizational policy subtleties, and the authority to resolve employee grievances, making them suitable only as drafting aids rather than reliable performers of the task.
Technical feasibility today	claude-sonnet-5	2/5	AI chat assistants can offer generic conflict-resolution scripts, but no deployed product reliably advises security supervisors in real-time on live, context-specific personnel and complaint situations.

Assign security personnel to posts or patrols.

CI 16–39 · exposure 17 · augmentation 50 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security services remain labor-intensive, fragmented across small and mid-sized firms with low digitization; adoption of AI-driven scheduling is nascent and limited mostly to large corporate or government entities.
Sector adoption velocity	claude-sonnet-5	2/5	Security services are a relatively low-digitization, physical-labor-heavy sector with slower AI adoption compared to information/finance industries.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI could usefully assist by flagging scheduling conflicts, suggesting optimizations, or recommending assignments based on historical patterns, allowing supervisors to make faster, more informed decisions while retaining control.
Augmentation potential	claude-sonnet-5	3/5	AI-based scheduling and optimization tools can meaningfully assist supervisors in efficiently allocating personnel, though final decisions remain human-driven.
Task automatability	claude-haiku-4-5-20251001	1/5	Assigning personnel to posts requires understanding real-time contextual factors (staffing availability, site-specific risks, personnel capabilities, schedule conflicts) and making judgment calls that integrate human knowledge. Current AI cannot reliably perform this end-to-end without substantial human oversight, falling well short of the 50% time-saving threshold.
Task automatability	claude-sonnet-5	2/5	Scheduling/assignment logic could be partially automated with rostering software, but real-time judgment about personnel fit, incidents, and site risk still requires human oversight, so full end-to-end automation isn't yet at the 50% threshold.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security operations often require supervisor accountability and judgment over staffing decisions; liability for inadequate coverage, safety implications of poor assignments, and organizational practice of human responsibility for roster decisions create substantial friction against full automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement mandates a human specifically for assignment tasks, though liability for security lapses and client expectations of human accountability create some friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Integration and oversight costs for an AI assignment system would be significant, and the loaded cost of a supervisor (who currently handles this in minutes as part of broader duties) is relatively low, making full automation economically marginal at best.
Cost vs. human wage	claude-sonnet-5	3/5	Scheduling software is cheap relative to a supervisor's wage, but the assignment task still requires human review and adjustment, keeping overall cost roughly comparable once oversight is included.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI could assist with scheduling logistics or flag conflict detection, no deployed product reliably handles the full task of personnel assignment in security contexts; existing workforce scheduling tools are narrow and require heavy manual curation.
Technical feasibility today	claude-sonnet-5	2/5	Workforce scheduling tools exist and are used in security firms, but they mostly support rather than autonomously execute post/patrol assignment decisions, especially under dynamic conditions.

Explain company policies and procedures to staff using oral or written communication.

CI 25–30 · exposure 25 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security sector shows below-average AI adoption relative to tech and finance; most firms still rely on supervisors for policy communication rather than automating or augmenting it with AI systems.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a lower-digitization sector with limited AI agent deployment for internal supervisory communications compared to information or finance sectors.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by generating clear policy drafts, scripts, FAQs, and written summaries that a supervisor then delivers or customizes; this moderately improves preparation and consistency, though the delivery and dialogue remain human-centered.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly help supervisors draft clear policy documents, talking points, and FAQs, improving consistency and speed while the supervisor still delivers the communication.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can generate written policy explanations and draft materials efficiently, but the task requires adaptively explaining policies to diverse staff with real-time Q&A, feedback interpretation, and organizational context—elements that demand human judgment and presence. Some preparation work is automatable; the core supervisory function is not.
Task automatability	claude-sonnet-5	2/5	AI can draft policy explanations but the actual act of explaining to staff, answering follow-up questions, and ensuring understanding requires human presence and authority, especially in supervisory contexts.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Supervisory authority and accountability for policy communication to employees carry implicit legal and organizational responsibility; employment law and liability expectations generally require a human supervisor to own policy communication, even if AI assists in drafting.
Adoption barriers	claude-sonnet-5	3/5	No licensing barrier specifically, but organizational expectation that supervisors personally communicate policy, plus liability for miscommunication of security procedures, creates moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI can reduce time spent drafting written materials, but the supervisor wage is relatively modest compared to knowledge work, and the actual task—interactive, in-person or synchronous communication—requires human presence that AI cannot replace cost-effectively.
Cost vs. human wage	claude-sonnet-5	2/5	While drafting text is cheap, the supervisory function (in-person clarification, tone-setting, authority) still requires paid human time, limiting cost savings for the full task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While LLMs excel at drafting policy documents and explanations, no deployed product reliably handles the interactive, context-sensitive aspects of explaining policies to live staff, answering clarifications, and adapting to staff comprehension levels in a supervisory role.
Technical feasibility today	claude-sonnet-5	2/5	AI chatbots and document generation tools exist for internal comms, but no deployed product reliably substitutes for a supervisor's direct explanation of policies to security staff in practice.

Train security personnel on protective procedures, first aid, fire safety, and other duties.

CI 25–29 · exposure 25 · augmentation 75 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While some security firms have piloted online training modules, adoption of fully AI-driven training remains limited. Most organizations still rely on in-person or live-instructor-led training, especially for hands-on skills and compliance documentation; production-scale autonomous security training is uncommon.
Sector adoption velocity	claude-sonnet-5	2/5	Security services is a lower-digitization, physically-oriented sector where AI adoption for training delivery remains nascent and pilot-stage.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly augment supervisor productivity by auto-generating training materials, creating adaptive learning paths, generating quizzes and assessment scaffolds, and tracking trainee progress—allowing supervisors to focus on live instruction, scenario drills, and competency validation. This is already happening in forward firms.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by generating training curricula, quizzes, scenario simulations, and tracking compliance, augmenting the trainer's efficiency significantly.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can assist with content creation (slides, manuals) and deliver online modules, but cannot reliably conduct hands-on training, observe worker competency, adapt to live questions, or evaluate individual performance—all essential to effective security personnel training. The task requires real-time feedback and scenario adaptation.
Task automatability	claude-sonnet-5	2/5	AI can generate training materials and quizzes but delivering hands-on protective procedure training, physical drills, and first aid certification requires human demonstration and supervision.5
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security training often requires certification, documented competency sign-off by a licensed supervisor, and liability accountability for failures in critical areas like first aid and fire safety. Regulatory and insurance frameworks typically mandate human instructor attestation and cannot be wholly displaced by AI systems.
Adoption barriers	claude-sonnet-5	4/5	First aid and fire safety training often require certified instructors and compliance with regulatory/accreditation standards, plus liability concerns limit full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI-driven LMS platforms and content generation can reduce per-trainee delivery costs compared to live instructor time, but integration, customization, and oversight still require human involvement. The cost savings are moderate rather than transformative, especially given compliance and certification requirements.
Cost vs. human wage	claude-sonnet-5	2/5	AI-generated content lowers material prep costs somewhat, but certified instructors and hands-on practice sessions still dominate cost, keeping overall savings modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Learning management systems (LMS) and AI-generated training content exist, but production systems for autonomous delivery of security training lack the ability to certify competency, handle practical demonstrations, or provide the personalized feedback that supervisors currently give. Most deployed solutions are content repositories, not autonomous trainers.
Technical feasibility today	claude-sonnet-5	2/5	E-learning platforms and chatbots exist for supplementary training content, but no deployed product reliably conducts full security/first aid/fire safety training programs end-to-end in production.

Patrol the premises to prevent or detect intrusion, protect property, or preserve order.

CI 16–30 · exposure 17 · augmentation 50 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of autonomous patrols remains limited, concentrated in high-security, controlled environments (data centers, warehouses). Most security firms still deploy human patrols; AI augmentation (monitoring dashboards, alert systems) is more common than replacement in real-world operations.
Sector adoption velocity	claude-sonnet-5	2/5	Security services are a relatively low-digitization physical sector; camera/AI monitoring is spreading but full patrol automation via robots remains niche and slow to scale.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist human patrols through real-time threat detection (camera feeds, anomaly alerts), historical analytics, and scheduling optimization. However, the human supervisor remains responsible for judgment and intervention, making augmentation helpful but not transformative in isolation.
Augmentation potential	claude-sonnet-5	3/5	AI-enhanced cameras, motion sensors, and analytics can flag anomalies and support supervisors in directing patrols more efficiently, improving situational awareness without replacing the physical patrol task.
Task automatability	claude-haiku-4-5-20251001	2/5	Physical patrols of premises require embodied presence, situational awareness, and judgment that current robots cannot reliably replicate end-to-end. While AI can assist with monitoring (cameras, alerts), autonomous systems cannot independently perform the full duty cycle of presence-based deterrence and intervention.
Task automatability	claude-sonnet-5	1/5	Physical patrolling of premises requires embodied presence and mobility that current AI cannot perform end-to-end; robots or drones remain narrow, expensive pilots, not general substitutes for human patrol.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong legal and liability barriers exist: security personnel often carry legal authority to detain, intervene, and use force; liability asymmetry is high (an AI incident can trigger significant damages). Premises protection also commonly involves client-specific authorization and regulatory compliance (licenses, insurance) tied to human personnel.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement for AI to patrol, but liability concerns, need for human judgment and response to incidents, and physical property considerations create meaningful friction against full substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Autonomous security systems (robots, advanced cameras) remain capital-intensive and require significant integration; human security workers remain cheaper per patrol-hour for broad, dynamic premises coverage. Full replacement cost per task equivalent remains comparable to or exceeds the loaded wage of entry-level security staff.
Cost vs. human wage	claude-sonnet-5	2/5	Autonomous patrol robots and sensor networks require significant capital investment, maintenance, and human oversight, making them not clearly cheaper than a security guard's wage in most settings.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Security robotics exist for limited surveillance and monitoring in controlled environments, but deployed products cannot reliably detect intrusion, assess threats, or preserve order across diverse premises. Most production deployments remain narrowly scoped complements to human patrols rather than replacements.
Technical feasibility today	claude-sonnet-5	2/5	Some deployed security robots and camera-based AI monitoring exist, but they operate alongside human patrols with narrow coverage and material limitations rather than replacing the patrol function itself.

Secure entrances and exits by locking doors and gates.

CI 7–30 · exposure 13 · augmentation 38 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security work remains concentrated in physical, on-site roles with lower digital maturity. Adoption of smart locking is uneven and slower in smaller facilities and high-liability sectors. Most security firms still rely on manual rounds and human verification, with automation limited to high-tech campuses or large enterprises.
Sector adoption velocity	claude-sonnet-5	2/5	Physical security and facilities management are relatively low-digitization sectors where automated locking systems are adopted slowly and unevenly, mostly in newer or high-security facilities.
Augmentation potential	claude-haiku-4-5-20251001	3/5	Smart locks and remote monitoring can assist supervisors by providing real-time alerts and remote unlock capability in emergencies, reducing time spent on physical rounds. However, the assistance is partial—human verification of security posture remains central to the role, so augmentation is meaningful but not transformative.
Augmentation potential	claude-sonnet-5	2/5	Smart access control and monitoring systems can alert supervisors to unlocked doors or breaches, offering some assistance, but do not fundamentally transform this physical task.
Task automatability	claude-haiku-4-5-20251001	2/5	Locking doors and gates requires physical manipulation in varied environments (different lock types, configurations, accessibility). While some smart locks can be controlled remotely, the task includes securing multiple entrances/exits that may lack automation, and verification that locks are properly engaged typically requires human judgment or visual inspection. Current AI cannot reliably perform the full end-to-end task.
Task automatability	claude-sonnet-5	1/5	This requires physical presence to manipulate locks, doors, and gates; no off-the-shelf AI system can perform the physical act of securing entrances end-to-end today.a
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security and access control are heavily regulated; premises liability and duty-of-care standards typically require a responsible human to verify entrances are secured. Many organizations mandate that a licensed or authorized human supervisor personally confirm locks are engaged, creating a legal and contractual barrier to full automation.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement blocks automation specifically, but liability for security breaches, physical infrastructure constraints, and reliance on human judgment for irregular situations create real friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Smart lock installation and integration costs are substantial per entrance, and oversight/maintenance add complexity. For a security supervisor earning ~$35–45k annually, the amortized cost of automating locking duties across a facility often remains comparable to or exceeds the labor savings, especially for older or non-standardized infrastructure.
Cost vs. human wage	claude-sonnet-5	1/5	Retrofitting facilities with automated locking/access systems plus monitoring is capital-intensive and often costs more than having existing security staff perform this simple physical task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Smart lock systems exist and can be remotely operated, but they represent a narrow portion of real-world security infrastructure. Many facilities use traditional locks, padlocks, or gates without automation capability. Deployed products do not reliably handle the full scope of entrance/exit securing across diverse facility types.
Technical feasibility today	claude-sonnet-5	1/5	While smart locks and automated access control systems exist, they require pre-installed infrastructure and don't autonomously perform the supervisory judgment and physical securing task as described in production at scale.

Investigate disturbances on the premises, such as security alarms, altercations, and suspicious activity.

CI 7–25 · exposure 13 · augmentation 50 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Security remains a heavily human-supervised field with slow digital transformation. Most firms use AI only for initial alarm classification and monitoring assist; actual investigation remains human-driven. Adoption is primarily in pilot phases for automation, not production-scale replacement in core investigation roles.
Sector adoption velocity	claude-sonnet-5	2/5	Security services sector is a low-digitization, physically-oriented field where AI adoption for investigative response remains nascent, mostly limited to camera/alarm monitoring tools.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist supervisors with alarm prioritization, historical pattern matching, and data review (logs, camera feeds), raising their efficiency in deciding where to investigate. However, the assistance is partial—the supervisor must still conduct the physical investigation and make judgment calls, so augmentation is useful but not transformative.
Augmentation potential	claude-sonnet-5	3/5	AI-enabled cameras, sensors, and alert systems can flag disturbances and provide data to guide the supervisor's investigation, improving situational awareness even though the physical investigation remains human-led.
Task automatability	claude-haiku-4-5-20251001	2/5	Investigation of disturbances requires contextual judgment, human interaction, and physical presence to assess complex situations (altercations, suspicious activity). While AI could assist with alarm classification and initial triage, the core investigative task—evaluating context, interviewing people, and deciding on appropriate response—requires human reasoning and cannot meet the 50% time-saving threshold end-to-end.
Task automatability	claude-sonnet-5	1/5	Investigating physical disturbances requires on-site presence, physical judgment, and real-time human interaction that current AI cannot perform end-to-end.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Significant barriers exist: liability exposure (if an AI-driven investigation misses a real threat, organizations face legal risk), regulatory requirements in many jurisdictions mandate human security authority presence, and customer/tenant expectations strongly favor human judgment in security matters. The asymmetric cost of errors (missed security incidents) creates high protection.
Adoption barriers	claude-sonnet-5	4/5	Security supervisory roles often require licensing, liability for use-of-force decisions, and physical authority that legally must rest with an accountable human.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	A first-line supervisor's loaded hourly cost is substantial ($30–50+/hour), while AI systems for alarm monitoring/triage cost far less per incident. However, the AI still cannot fully replace the investigative work, so the ratio remains unfavorable for full automation—human oversight is still required and expensive.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical, in-person investigative task, so a cost comparison favors the human who can actually perform it.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Current AI can assist with alarm detection/classification and log analysis, but no deployed product performs end-to-end disturbance investigation reliably. Security systems exist but still require human investigators for actual scene assessment, evidence gathering, and judgment calls about what constitutes a genuine threat.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously investigates physical security incidents; AI is limited to alerting/detection support, not investigation itself.

Inspect and adjust security equipment to ensure it is operational or to detect evidence of tampering.

CI 5–28 · exposure 13 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Security firms remain largely traditional, with heavy reliance on human rounds and manual inspection; digital and automated security monitoring adoption is growing but heavily concentrated in high-end corporate/government sectors, not yet widespread in typical security operations.
Sector adoption velocity	claude-sonnet-5	1/5	Security services is a physically-oriented, lower-digitization sector with limited AI agent deployment for hands-on equipment maintenance tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI tools such as automated log analysis, anomaly detection on sensor feeds, and computer-vision pre-screening of equipment can assist supervisors in prioritizing which equipment to inspect, but the core judgment and physical inspection remain human-led.
Augmentation potential	claude-sonnet-5	3/5	AI-enabled diagnostic software and remote monitoring systems can flag anomalies or tampering signals, helping supervisors prioritize physical inspections, though the core inspection and adjustment remains manual.
Task automatability	claude-haiku-4-5-20251001	2/5	Current AI systems can assist with visual inspection of equipment (via computer vision) and log analysis, but detecting subtle tampering, assessing operational status under varied conditions, and making contextual safety decisions requires human judgment and physical interaction that falls short of the 50% time-saving threshold today.
Task automatability	claude-sonnet-5	1/5	This requires physical inspection of hardware (cameras, sensors, locks, alarms) and hands-on adjustment, which current AI systems cannot perform without robotic embodiment.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security functions carry liability and legal compliance requirements; inspecting equipment and detecting tampering often requires a human agent authorized and accountable for safety decisions, and many jurisdictions require licensed security personnel to certify system integrity.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing barrier exists specifically for this inspection task, but liability for security failures and physical access requirements create organizational friction against automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Automated inspection hardware, vision systems, and integration labor are substantial upfront costs; the loaded wage of a security supervisor is moderate but spread across daily rounds, making the cost comparison unfavorable for automation at this task level.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute for the physical inspection/adjustment component, so AI cost is essentially irrelevant while the human cost remains necessary.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Specialized robotic systems and vision tools exist for narrow equipment checks, but no deployed product reliably inspects a full range of security equipment, detects tampering, and assesses operational integrity across diverse contexts in production environments.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product physically inspects and adjusts security equipment autonomously; this remains a manual field task performed by human supervisors.

Screen individuals and belongings to prevent passage of prohibited materials using walkthrough detectors, wands, or bag searches.

CI 5–25 · exposure 13 · augmentation 38 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While automated detection equipment is widely deployed, actual AI-driven autonomous screening remains rare in production. Most facilities still rely on human operators interpreting machine signals; adoption of fully autonomous systems is slow due to safety-critical nature and regulatory requirements.
Sector adoption velocity	claude-sonnet-5	1/5	Security services is a low-digitization, physically embodied sector where AI adoption for hands-on screening is minimal to nonexistent in production settings.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-enhanced detection systems (improved X-ray algorithms, anomaly flagging) can assist human screeners by highlighting suspicious items and reducing false positives. However, the human remains the primary decision-maker, making this partially augmentative rather than transformative.
Augmentation potential	claude-sonnet-5	2/5	AI-enhanced scanners and image-analysis software can assist in flagging suspicious items in X-ray images, offering some augmentation, but the core physical screening and judgment calls remain human-driven with limited AI integration reported here.
Task automatability	claude-haiku-4-5-20251001	2/5	While metal detectors and X-ray systems can identify some prohibited materials, the task requires human judgment to interpret ambiguous readings, assess context, and make final authorization decisions. Current AI cannot reliably replace the full end-to-end task including discretionary decision-making at the 50% time-saving threshold.
Task automatability	claude-sonnet-5	1/5	This is a hands-on physical screening task requiring physical manipulation of detectors, wands, and bags on individuals, which current AI cannot perform end-to-end without robotic embodiment far beyond deployed capability.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Security screening is heavily regulated (TSA, OSHA, facility-specific security policies) and often requires licensed or certified personnel. Legal liability for missed threats and errors creates strong organizational and regulatory friction against fully automated systems without human authorization.
Adoption barriers	claude-sonnet-5	4/5	Security screening often involves legal authority, physical contact with individuals, and liability/regulatory requirements (e.g., TSA-style certifications) that generally require a human presence and judgment, though not always formal licensure.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Detection equipment is capital-intensive and requires integration, calibration, and human monitoring. The total cost (hardware, maintenance, support staff) remains comparable to or exceeds the loaded wage of a single security screener, especially for lower-volume facilities.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this physical screening task, so no meaningful cost comparison favors AI; human labor remains the only functional option.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Automated detection equipment (metal detectors, X-ray machines) exists but requires human operators to interpret results and make final screening decisions. No deployed AI system independently performs the full screening and authorization task; human oversight remains mandatory and non-delegable.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously operates walkthrough detectors, wands, or conducts physical bag searches on people; this remains firmly in the domain of human security personnel.

Call police or fire departments in cases of emergency, such as fire, bomb threats, and presence of unauthorized persons.

CI 0–5 · exposure 5 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Security operations remain heavily dependent on human judgment and accountability; adoption of autonomous emergency-calling systems in production is essentially nonexistent due to legal and safety constraints.
Sector adoption velocity	claude-sonnet-5	1/5	Security services are a physical, low-digitization sector with minimal AI adoption for real-time emergency decision-making and communication tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by automatically detecting potential emergencies (fire, unauthorized entry), alerting supervisors, and providing relevant context, but the supervisor must remain in the decision loop to call emergency services.
Augmentation potential	claude-sonnet-5	3/5	AI-enabled sensors, alarm systems, and monitoring software can help detect anomalies and alert supervisors faster, augmenting situational awareness even though the call itself remains human-driven.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires real-time human judgment to assess emergency severity, decide whether police or fire response is appropriate, and communicate nuanced situational details—decisions that cannot be reliably automated end-to-end by current AI systems without human oversight that negates time savings.
Task automatability	claude-sonnet-5	1/5	This requires real-time human judgment, physical presence, situational awareness, and decisive action during emergencies; no AI system can independently detect and decide to call emergency services with accountability today.'
Adoption barriers	claude-haiku-4-5-20251001	5/5	Legal liability for false emergency calls, potential criminal penalties, and the requirement that a responsible person make the decision to contact emergency services create hard barriers; dispatchers and law enforcement expect a human on the line who can be held accountable.
Adoption barriers	claude-sonnet-5	5/5	Emergency response communication involves legal liability, chain-of-command authority, and often requires a human decision-maker who can be held accountable; automating this raises serious liability and safety concerns.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of AI systems that monitor for emergencies plus required human oversight and liability management would likely exceed the minimal cost of a human supervisor simply making a phone call when needed.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this task, so cost comparison favors the human by default since AI cannot reliably execute it at all.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can draft emergency call templates or flag potential emergencies, no deployed system reliably makes the decision to call emergency services autonomously and completes the full communication without human verification, given liability and legal requirements.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously makes emergency calls to police/fire departments in place of a human security supervisor; this remains firmly a human responsibility.

Apprehend or evict trespassers, rule violators, or other security threats from the premises.

CI 0–0 · exposure 0 · augmentation 25 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	No meaningful AI adoption for physical apprehension or eviction tasks exists in security sectors. The task remains purely human-performed due to legal requirements and the need for embodied intervention.
Sector adoption velocity	claude-sonnet-5	1/5	Physical security work is a low-digitization, hands-on sector with minimal AI agent deployment for physical confrontation tasks; adoption in this specific function is essentially nonexistent.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could provide limited assistance through threat detection, surveillance alerts, or suspect identification to inform a security worker's decision, but does not transform productivity on the core physical apprehension or eviction task itself.
Augmentation potential	claude-sonnet-5	2/5	AI can assist with detection (cameras, analytics flagging trespassers) that alerts a supervisor, but it offers minimal help with the actual physical act of apprehension or eviction.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires physical apprehension or eviction of people, which demands embodied presence, force application, legal authority, and real-time situational judgment that current AI systems cannot perform. No autonomous system can reliably and safely apprehend or physically evict individuals.
Task automatability	claude-sonnet-5	1/5	This requires physical presence, physical force capability, and real-time judgment to confront and remove people from a location—something no current AI system can execute end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Hard legal and regulatory barriers exist: security personnel must be licensed or certified, liability exposure is severe, use of force is heavily regulated, and only authorized humans can legally apprehend or evict individuals from premises.
Adoption barriers	claude-sonnet-5	5/5	Physical intervention, use of force, and legal authority to detain or evict are tightly bound to human judgment, liability, and often legal/licensing requirements (e.g., security guard licensing, use-of-force training), making this a hard barrier to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI systems cannot perform this task at all, making cost comparison moot. The task requires a trained human security worker with legal authority and physical capability.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing physical apprehension, so any AI cost comparison is moot—human labor with physical capability is required and cheaper than any hypothetical automated alternative.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product performs apprehension or physical eviction of trespassers. This task falls entirely outside the scope of current AI capabilities, which lack physical autonomy and legal standing to enforce removal.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product physically apprehends or evicts people; robotics for physical security intervention remains research-stage or extremely narrow (e.g., stationary surveillance robots that cannot forcibly remove anyone).

Related occupations — Protective Service

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.