Private Detectives and Investigators
33-9021.00Median wage $51,220/yr35,580 employed (US)Rank #453 of 923 scored · top 49% by substitution
Gather, analyze, compile, and report information regarding individuals or organizations to clients, or detect occurrences of unlawful acts or infractions of rules in private establishment.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution28
Exposure27
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
14 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
7%
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%26
panel mean rating 2.1/5 → substitution pressure 26/100
Technical feasibility todayw 20%27
panel mean rating 2.1/5 → substitution pressure 27/100
Cost vs. human wagew 15%32
panel mean rating 2.3/5 → substitution pressure 32/100
Adoption barriersw 20%inverted — strong barriers lower the score33
panel mean rating 3.7/5 (barrier strength) → substitution pressure 33/100
Sector adoption velocityw 10%23
panel mean rating 1.9/5 → substitution pressure 23/100
Task breakdown (14 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.
Count cash and review transactions, sales checks, or register tapes to verify amounts or to identify shortages.
72CI 67–76 · exposure 70 · augmentation 75 · importance 3.7/5 · click for rater detail
Count cash and review transactions, sales checks, or register tapes to verify amounts or to identify shortages.
72| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Retail, hospitality, and financial services have begun deploying AI-driven transaction analysis and reconciliation tools, but adoption remains unevenly distributed and often limited to larger organizations. Private investigation firms operate in smaller, less digitized cohorts with slower modernization patterns. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Retail and finance sectors have adopted automated reconciliation and fraud-detection tools broadly, but private investigation as a field remains a small, low-digitization niche with slower uptake. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI dramatically accelerates cash counting via image analysis and automates discrepancy flagging in transaction logs, allowing the investigator to focus on interpreting anomalies and pursuing leads rather than manual tabulation. This represents a substantial productivity multiplier while the human remains in analytical control. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-assisted anomaly detection and automated reconciliation can significantly speed up an investigator's ability to spot discrepancies, while the investigator retains judgment over interpreting findings and pursuing leads. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can reliably count cash via image analysis and extract transaction/register data from digital sources or scanned documents with high accuracy, achieving significant time savings. However, the physical cash-counting component still requires human presence or specialized hardware, preventing full end-to-end automation without some manual intervention. |
| Task automatability | claude-sonnet-5 | 4/5 | Verifying cash counts against transaction records and register tapes is a structured, rule-based reconciliation task well-suited to automation via POS data integration and simple auditing software or AI-based anomaly detection. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal barriers exist to automating transaction counting and cash verification—no licensing requirement mandates human sign-off. Organizational friction from audit/compliance policy preferences is modest; the main barrier is legacy process inertia rather than regulation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human perform basic reconciliation, though findings feeding into legal investigations may need human verification and testimony credibility. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Inference cost for image/document processing is negligible; AI solutions cost orders of magnitude less than paying a human investigator's hourly rate for routine verification work, even accounting for integration and oversight. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated transaction-matching and anomaly-detection tools can process large volumes of register data far more cheaply than an investigator manually recounting cash and reviewing tapes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Production systems exist for OCR-based transaction extraction and financial reconciliation; computer vision can count currency in images with documented accuracy >95%. Deployed products handle register tape digitization and transaction verification in retail/hospitality contexts, though real-world performance varies by data quality. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Retail loss-prevention and POS reconciliation software already automate much of this cross-checking in production, though private investigators often perform this manually as part of ad hoc fraud investigations without dedicated tooling. |
Search computer databases, credit reports, public records, tax or legal filings, or other resources to locate persons or to compile information for investigations.
62CI 51–74 · exposure 62 · augmentation 75 · importance 4.4/5 · click for rater detail
Search computer databases, credit reports, public records, tax or legal filings, or other resources to locate persons or to compile information for investigations.
62| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Private investigation and legal sectors are digitally mature; adoption of automated database search and reporting tools is already widespread in production, with evidence of displacement in routine record-location tasks. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Private investigation is a small, fragmented, low-digitization industry with slow technology adoption despite occasional use of online record-aggregation tools. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists investigators by rapid multi-source searching, summarization, and cross-referencing, allowing humans to focus on interpretation, inference, and case strategy rather than manual data gathering. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly speeds up preliminary searches, pattern matching across records, and summarizing large volumes of data, meaningfully boosting investigator productivity while they retain oversight and judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI can systematically search and compile data from multiple structured databases and public records with high speed and accuracy, easily meeting the 50% time-saving threshold. However, some unstructured or restricted-access sources may still require human navigation or interpretation. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can rapidly query and synthesize structured public records, credit data, and legal filings, but investigators still need to verify accuracy, resolve identity ambiguities, and integrate findings with judgment-based leads, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While database access often requires licensing and data privacy regulations (FCRA, state laws) apply, many public records and credit-report searches are already semi-automated in the industry. Organizational friction and need for human verification of sensitive findings provide moderate protection. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Some data sources require licensed access or have legal restrictions (FCRA compliance for credit reports, privacy law), and investigators may need certification, but the search itself isn't inherently restricted to licensed humans by law in most jurisdictions. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven database searches and record compilation cost pennies per query with minimal setup; the loaded human cost for equivalent searching and compilation is substantial, making AI an order of magnitude cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated database queries and AI summarization are far cheaper per record than manual searching, though licensing fees for proprietary databases and verification overhead keep it from being an order of magnitude cheaper in all cases. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (legal research platforms with AI, database-search tools, automated report aggregators) reliably perform database queries and data compilation at scale. Some edge cases around access control or data verification remain, preventing a full 5. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products exist (data aggregation platforms, OSINT tools, AI-assisted search) that perform partial record retrieval and synthesis, but comprehensive, reliable cross-database investigation with legal-grade accuracy is not yet a mature turnkey product. |
Write reports or case summaries to document investigations.
62CI 51–72 · exposure 62 · augmentation 75 · importance 4.8/5 · click for rater detail
Write reports or case summaries to document investigations.
62| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Private investigation is a relatively small, less digitized sector compared to legal tech and finance; adoption of AI writing tools is growing but still in the pilot-to-early-production phase in most firms. Law enforcement and larger agencies move faster, but the sector overall remains moderate in adoption velocity. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Private investigation is a small, fragmented, often solo-practitioner industry with lower digitization and AI tool adoption compared to larger professional services sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments investigator productivity by converting rough notes and evidence into polished case summaries in minutes, allowing investigators to focus on analysis and field work. The human remains in control of fact-checking and interpretation, making this a strong productivity multiplier without full automation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI writing tools can substantially speed up drafting, organizing, and formatting case narratives from investigator notes, letting the investigator focus on fact-checking and judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can draft comprehensive reports from investigation notes, structured data, and evidence summaries with minimal human revision needed. Current language models reliably organize facts, generate coherent summaries, and produce professionally formatted documents that meet the ≥50% time-saving threshold, though final review by the investigator remains prudent. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft structured summaries from provided facts, evidence logs, and notes, saving significant drafting time, but investigators must supply accurate raw findings and verify legal/factual accuracy before use.ed. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While investigators may have contractual or licensure obligations to personally sign off on findings, the actual report writing task itself carries no strict legal requirement that a licensed human author it. Client preference for human involvement and quality review provide some friction but do not legally prevent AI authorship of drafts. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Reports may be used in legal proceedings, requiring accuracy, chain-of-custody integrity, and investigator accountability/signature, creating moderate liability and professional-standard barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI inference for report generation costs pennies per document, while professional investigators earning $50–80k/year spend 5–10% of billable time on report writing. Even with oversight costs, AI-generated draft reports are at least 10× cheaper than human-written equivalents at equivalent quality. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Drafting a report via an AI assistant costs a small fraction of the billable hourly rate of a private investigator, though human review and editing time still adds cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed tools (general LLMs, case management software with AI integrations, legal writing assistants) already perform report writing at scale in law enforcement and private investigation firms with acceptable accuracy for non-litigation summaries. Some edge cases and complex legal language may require human refinement, but basic to intermediate report generation is production-grade today. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | General-purpose LLM products (e.g., ChatGPT, Copilot) are already used for report drafting across professional fields, but no investigator-specific deployed product reliably handles case summary writing with domain-specific formatting and evidentiary rigor at scale. |
Conduct personal background investigations, such as pre-employment checks, to obtain information about an individual's character, financial status, or personal history.
43CI 36–50 · exposure 42 · augmentation 75 · importance 4.4/5 · click for rater detail
Conduct personal background investigations, such as pre-employment checks, to obtain information about an individual's character, financial status, or personal history.
43| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Background check automation is moderately adopted in large corporations and staffing agencies using established platforms, but many smaller firms and specialized investigation contexts still rely on manual processes. Adoption is faster in HR/compliance functions but slower in higher-stakes or complex investigations. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | HR and background-screening industries have adopted automated data aggregation tools fairly widely, but investigative firms handling complex personal history checks still rely heavily on manual work, giving a middling adoption pace. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments investigator productivity by automating data gathering, cross-referencing records, and flagging anomalies, while investigators focus on interpreting findings, conducting interviews, and making judgment calls. This assistant role transforms efficiency without removing the human from the investigative loop. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up record searches, data aggregation, and report drafting, letting investigators focus on verification, interviews, and judgment, meaningfully boosting productivity. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can automate a significant portion of background investigation workflow—database searches, public record aggregation, and initial report compilation—but requires human judgment to verify information credibility, interpret nuanced findings, and investigate inconsistencies. The task cannot be fully automated end-to-end without human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can rapidly search and aggregate public records, social media, and databases, but synthesizing findings, verifying accuracy, resolving identity ambiguity, and judgment about relevance still require substantial human review, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong regulatory and liability barriers exist: the Fair Credit Reporting Act (FCRA) imposes legal requirements on how data is collected and reported, investigations may require licensed investigator credentials, and error costs in pre-employment decisions create significant legal exposure. Organizations typically require human review and authorization of findings. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Background checks are subject to FCRA and similar regulations requiring accuracy, disclosure, and sometimes licensed investigator sign-off, creating moderate compliance and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI-powered background check services cost significantly less than manual investigation per record, but must still account for human review, verification, and report finalization. Integration with existing workflows and the need for human sign-off keep costs roughly comparable to mid-level investigator labor. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated data-pulling tools substantially cut research time and are cheaper than manual record searches, but human verification, interviews, and judgment calls keep overall cost from being an order of magnitude lower than a human investigator. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Deployed products (background check services, OSINT platforms, data aggregation software) perform parts of this task reliably, but they often have gaps in coverage, outdated records, or false positives that require human validation. No single system reliably handles all investigation aspects without human intervention. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Background-check platforms (e.g., Checkr, HireRight-integrated tools) already use automated data aggregation and some AI matching in production, but error rates around identity matching and incomplete records mean human investigators still verify and interpret results. |
Expose fraudulent insurance claims or stolen funds.
25CI 25–25 · exposure 25 · augmentation 75 · importance 4.4/5 · click for rater detail
Expose fraudulent insurance claims or stolen funds.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Insurance companies and large financial institutions pilot fraud-detection AI, but most investigations remain human-led; adoption is concentrated in high-volume claim triage, not investigative closure. Smaller investigative firms and solo practitioners lag significantly in AI integration. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Insurance fraud analytics has some AI adoption for claims triage, but the private investigation profession itself remains largely low-tech and slow to adopt full AI-driven workflows for fieldwork. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at rapid case-file review, linking disparate data sources, flagging anomalies, and surfacing historical patterns—tasks that can dramatically accelerate an investigator's productivity. Human detectives working with AI-assisted leads and analytics can investigate more cases faster while maintaining judgment and legal rigor. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly help investigators by flagging suspicious claims, analyzing financial records, and organizing large datasets, meaningfully speeding up the investigative process even though humans still conduct the core investigation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can assist with pattern detection, anomaly flagging, and data aggregation (claims databases, financial records), but exposing fraud requires human judgment about intent, contextual evaluation of evidence, witness credibility assessment, and often legal/courtroom testimony. The full task remains fundamentally human-driven. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can flag anomalies and cross-reference data patterns, but exposing fraud requires physical surveillance, interviews, evidence gathering, and judgment calls that current systems cannot execute end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Insurance fraud investigations and financial crime exposure often require licensed private investigator credentials, legal standing in court proceedings, and adherence to evidence-chain standards. Liability and regulatory oversight create strong friction against pure automation; a human expert must typically sign off on findings. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Private investigators often require licensing, and evidence used in fraud cases or legal proceedings typically needs to be gathered and attested to by a credentialed human investigator for admissibility and liability reasons. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI fraud-detection tools have significant upfront licensing and integration costs, and human investigators must still review flagged cases, conduct interviews, and build evidentiary chains. The all-in cost is comparable to, or slightly below, a junior investigator's wage for a limited subset of pattern-matching work. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI-driven anomaly detection is cheap for initial screening, but the full investigative task still requires costly human labor for fieldwork, making overall cost comparable to or only modestly cheaper than human investigators. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Deployed products exist for fraud detection and flagging suspicious claims, but they generate high false-positive rates and require human investigators to validate findings. No current system can independently expose fraud at the evidentiary standard needed for legal action without investigator review. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Fraud-detection analytics products exist and are used by insurers to flag suspicious claims, but the actual investigative exposure (surveillance, interviews, legal evidence collection) is not performed by deployed AI products today. |
Obtain and analyze information on suspects, crimes, or disturbances to solve cases, to identify criminal activity, or to gather information for court cases.
25CI 25–25 · exposure 25 · augmentation 75 · importance 4.3/5 · click for rater detail
Obtain and analyze information on suspects, crimes, or disturbances to solve cases, to identify criminal activity, or to gather information for court cases.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | While some investigative firms experiment with AI-assisted tools for preliminary research, the sector remains dominated by traditional human investigative practices. Regulatory requirements, need for courtroom credibility, and liability concerns slow production adoption despite growing availability of supporting technologies. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | The private investigation industry is small, fragmented, and slow to adopt advanced AI tools beyond basic database and search software, reflecting a low-digitization, laggard sector pattern. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at augmenting investigators by rapidly processing public records, generating leads from large datasets, identifying patterns in communication logs, and organizing evidence. These tools can substantially raise investigator productivity while the human retains judgment over case direction and evidentiary interpretation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly aids investigators by rapidly searching public records, social media, and databases, and by helping organize and cross-reference large volumes of information, meaningfully boosting productivity while humans retain judgment and legal responsibility. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data gathering, analysis of public records, and pattern recognition in large datasets, the task requires judgment about suspect credibility, contextual reasoning about criminal behavior, and integration of disparate sources that current AI systems handle inconsistently. The investigative work involves human-level inference and decision-making that AI cannot reliably perform end-to-end at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with data aggregation, records searches, and pattern analysis, but the core investigative work—surveillance, interviewing, judgment calls on credibility, and physical evidence gathering—requires human presence and reasoning that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Investigation work in court cases involves strict evidentiary standards, chain-of-custody requirements, and legal admissibility rules that create regulatory barriers. Private investigators must often be licensed, and courts require human verification and testimony; liability for incorrect findings is substantial and falls on the investigator. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Private investigators often require state licensing, and evidence used in court must meet chain-of-custody and admissibility standards, creating strong legal and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for records analysis and data gathering are relatively inexpensive, but the investigator's loaded wage remains competitive with the cost of integrating multiple AI systems, human review of outputs, and oversight necessary to ensure legal and evidentiary standards are met. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools reduce time spent on records searches and data correlation, but the overall investigative process still requires substantial human labor, oversight, and legal accountability, keeping costs comparable to or only modestly below human-only investigation. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some AI products exist for document analysis and database searching, but no deployed system reliably performs the full investigation task independently. Products that assist with OSINT and record analysis are available but narrow in scope and require substantial human oversight and interpretation of findings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist for OSINT aggregation, background checks, and link analysis, but no deployed system reliably conducts full investigations; these tools serve as narrow research aids rather than autonomous investigators. |
Investigate companies' financial standings, or locate funds stolen by embezzlers, using accounting skills.
25CI 25–25 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail
Investigate companies' financial standings, or locate funds stolen by embezzlers, using accounting skills.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Private investigation and forensic accounting remain traditional, relationship-driven, and heavily regulated sectors with slow digitization. While some firms use analytics tools, few organizations have moved to AI-driven investigation workflows, and cultural reliance on credentialed human investigators slows adoption. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Private investigation and forensic accounting are niche, low-digitization fields with limited large-scale AI deployment compared to fast-moving sectors like finance or tech; adoption is mostly limited to analytic tool use rather than transformative AI workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist investigators by automating document review, flagging statistical anomalies in financial data, and organizing voluminous records, raising efficiency. However, the human investigator remains essential for case strategy, source development, and evidence interpretation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully speed up transaction analysis, flag suspicious patterns, and assist with document review, substantially aiding investigators even though it doesn't replace the overall detective work. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data analysis and pattern detection in financial records, the task requires investigative judgment, witness interviews, and contextual reasoning about complex fraud schemes that remain firmly in human domain. Only narrow components (data aggregation, anomaly flagging) are automatable; the investigative synthesis requires human expertise. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with data analysis and pattern detection in financial records, but the core investigation requires physically obtaining records, interviewing sources, chasing leads across jurisdictions, and making judgment calls that current systems cannot execute autonomously end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Investigation results often serve as evidence in legal proceedings, requiring human accountability, expert testimony qualification, and compliance with evidentiary standards. Additionally, access to financial records is restricted and often requires licensed investigators with legal authority to conduct inquiries, creating significant regulatory and liability barriers. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Findings often need to hold up in court or regulatory proceedings, requiring licensed investigators or certified forensic accountants to testify and sign off, creating strong legal and liability-driven barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI tools for financial analysis are moderately priced, but integration into a complete investigation workflow, plus required human oversight and legal vetting, makes the total cost comparable to or higher than hiring experienced financial investigators and forensic accountants. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can cheaply flag anomalies in structured financial data, but the investigation still requires expensive human oversight, subpoenas, interviews, and legal proceedings, keeping overall cost comparable to or only modestly below human-only costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed AI system reliably performs financial forensics or embezzlement investigation end-to-end. Tools exist for data analysis and anomaly detection, but no production system integrates financial investigation with legal evidence standards and fraud case building required by the task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Forensic accounting software and AI-driven anomaly detection tools exist and are used by analysts, but no deployed product independently conducts full embezzlement investigations or financial standing assessments reliably in production. |
Alert appropriate personnel to suspects' locations.
20CI 10–30 · exposure 13 · augmentation 38 · importance 3.8/5 · click for rater detail
Alert appropriate personnel to suspects' locations.
20| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Private investigation remains a fragmented, localized sector with many small firms; digital adoption is slower than finance or tech. While surveillance automation is increasing, intelligent alerting workflows are still largely manual and case-dependent, limiting velocity. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Private investigation is a small, physically-oriented, low-digitization sector with minimal AI agent adoption for active field-based suspect tracking and alerting. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by monitoring feeds, flagging suspect locations in real time, and organizing alerts for the investigator to review and action. This augmentation is useful but limited by the need for human judgment on escalation and authorization, making it a partial productivity boost rather than transformative. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with communication tools, mapping, or notification systems to relay information faster, but it does not meaningfully enhance the core judgment-based act of identifying and alerting on a suspect's location. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can monitor surveillance feeds and flag anomalies, alerting appropriate personnel requires real-time location accuracy, authorization verification, and judgment about which personnel to contact in context-specific scenarios—tasks current systems handle inconsistently. The human operator must still interpret alerts and decide on action, preventing ≥50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires real-time physical surveillance, situational judgment, and immediate human communication in dynamic field conditions that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Private investigation operates under state licensing and may involve sensitive information requiring chain-of-custody or client approval for alert automation. Some jurisdictions have privacy and liability constraints, but no hard requirement mandates human alerting—moderate friction exists rather than legal prohibition. |
| Adoption barriers | claude-sonnet-5 | 3/5 | While not strictly licensed to a specific person for this micro-task, private investigation work often involves licensing requirements and liability concerns around surveillance and reporting suspect locations, especially involving law enforcement coordination. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-based location monitoring and alert systems require infrastructure (cameras, servers, integration with communication systems) and ongoing human review to validate alerts before personnel are notified. Total cost is comparable to or higher than a human investigator performing selective monitoring and targeted alerts. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical detection and alerting function, so cost comparison favors the human investigator by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Surveillance and anomaly-detection systems exist in production, but alerting the right personnel involves context-dependent authorization and escalation rules that vary by jurisdiction and case sensitivity. Current deployed systems require substantial human oversight to avoid false alarms and ensure alerts reach appropriate personnel correctly. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product tracks and alerts personnel about a physical suspect's location autonomously; this remains a human field operation with possibly some sensor/communication tools. |
Conduct private investigations on a paid basis.
15CI 5–25 · exposure 13 · augmentation 50 · importance 4.5/5 · click for rater detail
Conduct private investigations on a paid basis.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Private investigation remains a human-intensive, physically distributed sector with many small firms. While technology adoption is occurring (databases, digital tools), actual automation of investigation decisions remains limited. The sector is not experiencing rapid AI-driven displacement comparable to information-intensive fields. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | This is a small, physically-oriented, low-digitization occupation with little evidence of AI agent deployment replacing fieldwork or client-facing investigative services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI offers meaningful assistance on subtasks: automating OSINT, organizing data, flagging patterns in records, and drafting reports. These augment an investigator's productivity, but the human investigator must remain central to evidence evaluation, strategy decisions, and client-facing work. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with background research, data aggregation, record searches, and report drafting, meaningfully boosting productivity on the research portions of investigative work. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with some components—data aggregation, background checks, pattern analysis—the task fundamentally requires field work, human judgment, witness interviews, and adaptive decision-making under uncertainty that current AI cannot autonomously execute end-to-end. Most of the core investigative work remains dependent on human operators. |
| Task automatability | claude-sonnet-5 | 1/5 | This task is a broad, high-level umbrella covering surveillance, interviews, evidence gathering, and physical presence in the world—none of which current AI can perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: private investigator licensing is legally required in most jurisdictions, many tasks require a licensed human to perform them directly, liability for investigative errors is asymmetric, and client relationships demand human accountability. Regulatory frameworks explicitly designate who can conduct investigations legally. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Private investigators typically require state licensing, and their work carries legal/liability implications (evidence admissibility, privacy law), creating strong regulatory and professional barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for background checks and data analysis are relatively cheap, but deploying a human investigator remains necessary for most investigative work. The total cost including AI oversight, case management, and human investigator time is not materially cheaper than hiring an investigator outright. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot perform the physical and interpersonal components of investigations, so there is no viable AI-only cost comparison; a human investigator remains necessary for the core deliverable. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some products exist for specific subtasks (e.g., OSINT tools, database searches, report generation), but no deployed AI system reliably performs the full investigation task autonomously. Products like investigative software assist humans but do not replace the investigator's judgment, client interaction, or field presence. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts private investigations autonomously; investigations require physical surveillance, human judgment, and legal navigation that remain research-stage or nonexistent for AI. |
Question persons to obtain evidence for cases of divorce, child custody, or missing persons or information about individuals' character or financial status.
15CI 5–25 · exposure 13 · augmentation 50 · importance 4.3/5 · click for rater detail
Question persons to obtain evidence for cases of divorce, child custody, or missing persons or information about individuals' character or financial status.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Private investigation remains a traditional, human-relationship-dependent field with limited digital transformation. Most adoption is in backend analysis (databases, record searches) rather than the core task of interviewing subjects. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Private investigation is a small, low-digitization field with minimal AI agent deployment for direct human interviewing tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist investigators by summarizing interview transcripts, flagging inconsistencies in statements, organizing background information, and suggesting follow-up questions—useful support but the investigator remains essential to the questioning process itself. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help investigators prepare interview questions, organize notes, transcribe conversations, and cross-reference background data, but it doesn't perform the core interpersonal task. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can help analyze responses and generate interview templates, the core task—questioning people to assess credibility, detect deception, and build rapport—requires human judgment, empathy, and adaptive real-time interaction that current AI cannot reliably perform end-to-end with significant time savings. |
| Task automatability | claude-sonnet-5 | 1/5 | Interviewing people to obtain sensitive evidence requires in-person rapport, reading nonverbal cues, adaptive follow-up questioning, and legal judgment about admissibility—capabilities current AI cannot autonomously perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Legal admissibility of evidence, liability for investigative findings, client confidentiality, and the requirement for human testimony or credible human judgment in court proceedings create significant regulatory and liability barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Many jurisdictions require licensed investigators for such work, and evidence gathered improperly can be inadmissible or create liability, creating strong professional and legal barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI tools (transcription, document analysis) reduce some administrative overhead, but investigative questioning still requires trained human investigators. The fully-loaded cost of human investigators remains comparable to or lower than the cost of AI systems plus human oversight needed to validate findings. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this function, so cost comparison favors the human investigator by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs investigative questioning autonomously. AI chatbots exist but cannot substitute for a human investigator conducting sensitive interviews on divorce, custody, or character assessments in legal contexts where credibility assessment is critical. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts investigative interrogations or interviews for legal/personal cases; this remains purely a human field skill. |
Observe and document activities of individuals to detect unlawful acts or to obtain evidence for cases, using binoculars and still or video cameras.
15CI 5–25 · exposure 13 · augmentation 50 · importance 4.3/5 · click for rater detail
Observe and document activities of individuals to detect unlawful acts or to obtain evidence for cases, using binoculars and still or video cameras.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Private investigation is a small, specialized sector with limited digital infrastructure, high reliance on human judgment and courtroom credibility, and no measurable production-scale adoption of autonomous surveillance AI; adoption remains in pilot or tool-assisted mode. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Private investigation is a small, low-digitization field with minimal AI agent deployment for core fieldwork tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-powered video analysis (automated motion detection, object recognition, timeline assembly) can meaningfully assist investigators in reviewing footage and organizing findings, but the investigator must remain the decision-maker on what constitutes evidence and how to interpret observations legally. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with reviewing recorded footage, flagging events of interest, or organizing case documentation, but the core observational task remains human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Video recording and continuous monitoring can be partially automated with camera systems, but the core task of detecting meaningful unlawful acts requires contextual judgment, legal knowledge of what constitutes evidence, and adaptive decision-making about what to observe and document that current AI systems cannot reliably perform at production quality. |
| Task automatability | claude-sonnet-5 | 1/5 | Physical surveillance requires mobility, real-time visual judgment, discretion, and adaptive following in unpredictable environments—capabilities current AI systems lack outside narrow fixed-camera contexts. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Evidence gathered for legal proceedings must often meet strict admissibility standards and chain-of-custody requirements; many jurisdictions require a licensed investigator to collect, document, and testify about evidence, creating regulatory and legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Legal admissibility of evidence, licensing requirements for investigators in many jurisdictions, and chain-of-custody concerns create strong barriers to non-human execution of this task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | While automated video capture is cheap, the oversight, interpretation, and legal validation required still demand skilled investigators; the all-in cost of AI systems plus mandatory human review approaches or exceeds the loaded wage of an investigator. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute for physically tailing and observing a subject, so cost comparison favors human investigators by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | AI can assist with video analysis (object detection, motion tracking) but deployed products do not reliably identify unlawful acts or determine evidentiary value autonomously; human review and expertise remain essential for legal and investigative accuracy. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously conducts mobile human surveillance operations; AI is at most used for post-hoc video analysis, not live field observation and evidence gathering. |
Confer with establishment officials, security departments, police, or postal officials to identify problems, provide information, or receive instructions.
9CI 5–14 · exposure 5 · augmentation 38 · importance 3.6/5 · click for rater detail
Confer with establishment officials, security departments, police, or postal officials to identify problems, provide information, or receive instructions.
9| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | These inter-institutional conversations are conducted by private investigators themselves in established professional networks. No sector data shows momentum toward AI conducting these official liaisons; they remain human-mediated by necessity and professional norm. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Private investigation is a small, low-digitization field with in-person and phone-based liaison work, showing minimal AI adoption for this specific interpersonal function. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by preparing background briefings, drafting talking points, or summarizing meeting notes, but the core task of conferring with officials is inherently relational and depends on the investigator's presence and credibility, limiting augmentation scope. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help investigators prepare talking points, organize case information, draft follow-up summaries, and manage communication logs, meaningfully aiding but not replacing the human conferring itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task is fundamentally interpersonal and discretionary, requiring real-time dialogue, relationship management, and contextual judgment to navigate institutional hierarchies and extract or convey sensitive information. Current AI cannot reliably conduct these nuanced conversations without human oversight. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires real-time interpersonal conferring, building trust and rapport with officials, and situationally adapting communication, which current AI cannot substitute for end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Institutional protocols typically require verified human identity and professional credentials; police and postal officials are unlikely to share sensitive investigative information with an AI system rather than a licensed investigator. Legal and procedural norms create strong friction against automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Interactions with police and officials often require identity verification, legal authority, confidentiality, and trust relationships that effectively require a human licensed investigator, though not a strict licensing signature requirement. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI could support preparation of conference agendas or summarization of outcomes, but the core activity—meeting with officials—still requires a human investigator's time. Any cost savings are marginal because the human task duration is not substantially reduced. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this interpersonal liaison function, so no meaningful cost comparison favors AI; a human must be present and paid regardless. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product independently conducts official conferences with law enforcement or institutional authorities. While chatbots can draft communications or summarize information, actually meeting with officials and negotiating investigative cooperation requires human presence and credibility that AI cannot substitute. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts autonomous liaison meetings with police, security, or postal officials on behalf of an investigator; this remains firmly human-executed. |
Perform undercover operations, such as evaluating the performance or honesty of employees by posing as customers or employees.
3CI 0–5 · exposure 0 · augmentation 25 · importance 3.9/5 · click for rater detail
Perform undercover operations, such as evaluating the performance or honesty of employees by posing as customers or employees.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task inherently requires human field presence and cannot be displaced by information-era automation. Adoption of AI in investigation is limited to back-office analytics and research, not fieldwork substitution. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Private investigation is a small, physically-embedded, low-digitization sector with essentially no AI agent deployment for this specific undercover function. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can modestly assist by analyzing prior intelligence, flagging behavioral patterns, or coaching preparation, but the core undercover task remains human-dependent with limited scope for augmentation tools. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could help with case prep, background research, or generating cover-story materials, but offers minimal assistance during the actual undercover interaction itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Undercover operations require physical presence, social interaction, real-time behavioral adaptation, and deception credibility in varied, unscripted human contexts—capabilities fundamentally unavailable to current AI systems. No meaningful automation is feasible for the core task. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires embodied, real-world physical presence, human social deception, and in-person interaction that no current AI system can replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Undercover work is legally and contractually constrained: impersonation laws, trespass restrictions, employment law compliance, and organized-labor protections all require human judgment and legal accountability. Many jurisdictions require licensed investigators to perform this work. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Undercover work often requires licensing, legal authorization, and human judgment/liability considerations, plus the fundamental requirement of physical human presence. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task requires human presence and credibility; there is no AI substitute. AI support tools (surveillance systems, data analysis) are supplementary only, and human investigators remain the primary cost driver. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot perform this physical, in-person task at all, so there is no viable AI cost basis to compare against human labor. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product can perform undercover fieldwork, customer interaction, or embodied infiltration roles. The task requires physical agency and sustained human-like presence in real organizational environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs undercover physical investigation posing as a customer or employee; this is entirely outside current AI product capability. |
Testify at hearings or court trials to present evidence.
0CI 0–0 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail
Testify at hearings or court trials to present evidence.
0| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | There is no adoption to measure because automation is legally and constitutionally impossible. Testimony remains a core human legal function across all jurisdictions. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Courts and legal proceedings are highly conservative, rule-bound environments with essentially no movement toward AI performing witness testimony. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI might assist a detective in preparing testimony by organizing evidence, summarizing case details, or anticipating cross-examination questions, but such assistance is peripheral to the core act of testifying itself. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help investigators organize evidence, prepare summaries, and rehearse testimony beforehand, but it does not assist during the act of testifying itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Testifying at hearings or court trials requires legal standing, credibility assessment, cross-examination response, and real-time judgment under oath—tasks that are fundamentally tied to human legal personhood and accountability. No AI system can currently serve as a legal witness or substitute for testimony. |
| Task automatability | claude-sonnet-5 | 1/5 | Testimony requires a live, accountable human witness under oath who can be cross-examined; no AI system can perform this function today, even partially. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Extremely high barriers exist: only a human who was present and has personal knowledge can legally testify, and the witness must be sworn under oath and subject to cross-examination and perjury penalties. Legal rules of evidence and court procedure make substitution impossible. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Testimony is a legal act requiring a sworn, identifiable human witness subject to perjury laws, cross-examination, and courtroom rules—an absolute legal barrier to AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | This dimension is moot since the task cannot be automated at all; a human detective must personally testify, making any AI cost comparison irrelevant. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute cost to compare; the task cannot be performed by AI, so cost comparison is moot and defaults to AI being non-viable/more 'expensive' in effect. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can perform courtroom testimony; the legal system requires human witnesses with legal responsibility and the ability to be sworn under oath. AI testimony is not accepted in court proceedings. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product testifies in court in place of a human; this remains entirely outside current AI product scope. |
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.