Financial Examiners
13-2061.00Median wage $94,160/yr67,830 employed (US)Rank #516 of 923 scored · top 56% by substitution
Enforce or ensure compliance with laws and regulations governing financial and securities institutions and financial and real estate transactions. May examine, verify, or authenticate records.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution26
Exposure26
Augmentation63
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
17 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
0%
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%23
panel mean rating 1.9/5 → substitution pressure 23/100
Cost vs. human wagew 15%28
panel mean rating 2.1/5 → substitution pressure 28/100
Adoption barriersw 20%inverted — strong barriers lower the score23
panel mean rating 4.1/5 (barrier strength) → substitution pressure 23/100
Sector adoption velocityw 10%34
panel mean rating 2.4/5 → substitution pressure 34/100
Task breakdown (17 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.
Review balance sheets, operating income and expense accounts, and loan documentation to confirm institution assets and liabilities.
61CI 40–82 · exposure 66 · augmentation 88 · importance 4.1/5 · click for rater detail
Review balance sheets, operating income and expense accounts, and loan documentation to confirm institution assets and liabilities.
61| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services sector is among the fastest AI adopters; automated document review and compliance-checking systems are in production across banking and insurance. Pilot adoption is widespread, and displacement of routine review work is already measurable. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show strong AI adoption in analytics and fraud detection, but regulatory examination functions specifically adopt more cautiously due to compliance and audit trail requirements. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI significantly augments examiners by automating data extraction, flagging anomalies, and cross-validating entries, freeing the human to focus on judgment-heavy interpretations and institutional risk assessment. This is a canonical augmentation scenario in financial compliance. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids examiners by flagging anomalies, extracting data from lengthy loan documents, and speeding up initial review, while the examiner retains final judgment and sign-off. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Structured financial document review—balance sheets, income statements, and loan files—is highly automatable. Modern AI systems can parse, extract, and cross-validate numeric data and documentary evidence at scale with high accuracy, achieving ≥50% time savings over manual review. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can parse and cross-check financial statements and loan documents against records efficiently, but confirming asset/liability accuracy often requires judgment calls on valuation, context, and regulatory interpretation that current systems cannot fully replicate.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Regulatory expectation (e.g., bank regulators, audit standards) typically requires a licensed examiner to sign off on findings, and institutional risk tolerance for errors in asset-liability confirmation creates organizational friction. However, these are procedural/oversight barriers rather than hard legal prohibitions on automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examination is a regulated function often requiring credentialed examiners and formal sign-off, with significant liability for errors, creating strong barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Inference cost for document parsing and validation is negligible relative to loaded labor cost of a financial examiner. Even accounting for integration and oversight, AI-assisted review is an order of magnitude cheaper per task-equivalent. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI tools reduce time spent on document review and reconciliation, but the need for human oversight, verification, and liability coverage keeps blended costs closer to comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products (document AI, RPA platforms, accounting software) reliably extract and validate financial data in production environments. Some residual gaps remain in edge-case interpretations and complex multi-document reconciliation, but core functionality is deployable and widely used. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some fintech/RegTech products offer document extraction and anomaly detection for financial statements, but full end-to-end confirmation of institutional assets/liabilities in production examiner workflows remains narrow and human-supervised. |
Provide regulatory compliance training to employees.
49CI 36–62 · exposure 45 · augmentation 75 · importance 3.6/5 · click for rater detail
Provide regulatory compliance training to employees.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Financial institutions are adopting AI-assisted compliance training in pilots and early production (especially large banks), but sector-wide uptake remains moderate. Regulatory conservatism, liability concerns, and the need for human verification of content accuracy slow velocity compared to less regulated industries. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services is a fast-adopting sector for AI tools generally, but specifically for compliance training, adoption is moderate with pilots in content generation more common than full automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists compliance trainers and program managers by drafting content, personalizing modules by role/risk, and automating administration and assessment tracking. Human trainers remain in the loop for strategy and validation, while AI dramatically reduces manual content creation and delivery overhead. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly assist by drafting training content, generating scenario-based examples, updating materials for regulatory changes, and creating quizzes, greatly boosting trainer productivity. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can generate, deliver, and tailor regulatory compliance training content at scale with minimal human intervention. LLMs can create training modules, quizzes, and scenarios; learning management systems can automate delivery and tracking, achieving well over 50% time savings compared to manual instructor-led development and delivery. |
| Task automatability | claude-sonnet-5 | 2/5 | Content creation (materials, quizzes) can be AI-assisted, but live delivery, adapting to trainee questions, and certifying comprehension in a regulated context still require human involvement, so end-to-end automation is limited.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Regulatory bodies and internal compliance officers often require human sign-off on training accuracy and legal sufficiency, creating oversight friction. However, no hard licensing requirement mandates a human deliver the training itself, only that it be accurate and complete. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for delivering training itself, but organizations often require compliance officers or certified trainers to ensure regulatory accuracy and accountability, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven training generation and delivery is substantially cheaper than instructor-led or custom-built programs: one-time content creation via LLM plus LMS hosting costs far less than ongoing trainer salaries and development cycles. The cost advantage is typically 5–10× for large-scale rollouts. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply generate training materials and quizzes, but human trainers, compliance officers, and oversight for accuracy and regulatory nuance still add substantial cost, keeping overall savings moderate. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Deployed platforms (e.g., generative AI + LMS integration) can produce and deliver compliance training at scale, but accuracy on complex, jurisdiction-specific regulations remains inconsistent. Organizations still require human review of AI-generated content to catch errors, making purely autonomous deployment risky in high-stakes compliance contexts. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI-based e-learning platforms and chatbots exist for compliance training content, but reliable, regulator-accepted training programs delivered fully by AI at scale are not standard practice yet. |
Review and analyze new, proposed, or revised laws, regulations, policies, and procedures to interpret their meaning and determine their impact.
37CI 25–50 · exposure 38 · augmentation 75 · importance 3.7/5 · click for rater detail
Review and analyze new, proposed, or revised laws, regulations, policies, and procedures to interpret their meaning and determine their impact.
37| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services have adopted AI for document processing and initial screening, but core regulatory interpretation remains cautious and human-led. Adoption of autonomous interpretation systems in regulated compliance contexts is slow due to legal and operational risk aversion. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services is a fast-adopting sector for AI generally, but regulatory interpretation specifically is still in pilot/augmentation stages rather than widespread production reliance. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI document summarization, cross-referencing, and change-tracking significantly assist examiners in reviewing large volumes of regulatory updates and identifying likely impact areas, allowing them to focus expertise on judgment-heavy interpretation work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up scanning, summarizing, and comparing regulatory texts, letting examiners focus on judgment and impact assessment while staying in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can summarize and extract key points from regulatory documents, interpreting meaning and determining nuanced impact requires contextual judgment, understanding of edge cases, and synthesis with existing frameworks that current systems handle inconsistently. Automated interpretation alone cannot meet the 50% time-saving threshold without substantial human review and correction. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can summarize and flag changes in regulatory text quickly, but authoritative interpretation of legal impact for compliance decisions still requires expert judgment and accountability, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial examination is a regulated profession; examiners must be certified and their interpretations carry legal weight in compliance decisions. Liability for incorrect regulatory interpretation is high, and banking regulators expect human accountability, creating strong friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement mandates a human perform this specific analytical task, but institutional liability and regulatory expectations create moderate oversight friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI document processing is inexpensive per document, but the labor cost of human review to validate and refine interpretations still dominates the total cost. Oversight of AI output by experienced examiners means per-task cost remains comparable to or higher than hiring skilled analysts directly. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cut research time substantially, but the need for expert verification and liability review keeps overall cost roughly comparable rather than an order of magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Document analysis and summarization tools exist and are deployed, but reliable interpretation of regulatory impact—especially novel or ambiguous provisions—remains largely manual. AI products assist with retrieval and initial flagging but cannot independently perform impact determination at the quality required for financial regulation. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Legal/regulatory NLP tools and LLM-based compliance products exist and are used to summarize and compare regulations, but accuracy on nuanced legal interpretation remains inconsistent and requires human review. |
Review audit reports of internal and external auditors to monitor adequacy of scope of reports or to discover specific weaknesses in internal routines.
32CI 25–40 · exposure 33 · augmentation 63 · importance 4.0/5 · click for rater detail
Review audit reports of internal and external auditors to monitor adequacy of scope of reports or to discover specific weaknesses in internal routines.
32| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial regulation and audit governance tend to be conservative; while pilot projects exist, most regulatory bodies and audit firms have not deployed autonomous systems for scope/weakness determination, preferring human-led review with incremental AI assistance. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services and audit-adjacent fields are adopting AI at a moderate pace, with pilots for document analysis and anomaly detection common but full-scale deployment for judgment-heavy review still limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist by automatically extracting key audit procedures, flagging statistical outliers, cross-referencing prior findings, and organizing reports into comparative summaries, meaningfully reducing manual document review time while examiners retain final judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can significantly speed up review of large audit report volumes, highlight anomalies, and summarize key sections, meaningfully boosting examiner productivity while the human retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can extract and categorize information from audit reports and flag potential weaknesses through pattern matching, the task requires nuanced judgment about audit scope adequacy and systemic interpretation of internal control weaknesses that current systems struggle with. This is more assistive triage than autonomous end-to-end performance. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can extract, summarize, and flag anomalies across audit reports quickly, but judging adequacy of scope and materiality still requires professional judgment and contextual knowledge, limiting full end-to-end automation. Roughly half the review work (extraction, cross-referencing, flagging inconsistencies) can be automated today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial examination is subject to regulatory requirements (SEC, banking regulators) that typically require licensed examiners or qualified auditors to sign off on adequacy determinations and risk findings; liability for missed control weaknesses creates strong institutional resistance to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examiners are often licensed/regulated and final determinations of audit sufficiency and internal control weaknesses can carry legal and regulatory accountability, requiring sign-off by a qualified human examiner. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The oversight burden for AI-assisted review is substantial; examiners must still verify AI findings against complex audit evidence, and the cost of missed weaknesses is high. AI inference is cheap but total cost remains comparable to traditional human review due to verification overhead. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-assisted document review can cut analyst hours substantially, but the need for expert oversight, licensing, and error-checking keeps blended cost roughly comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some audit analytics products exist to scan reports and identify anomalies, but deployed systems typically operate with high false-positive rates and require substantial human validation before determining actual scope gaps or control weaknesses. Production use remains limited and heavily supervised. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some document-review and anomaly-detection tools are used in compliance/audit workflows, but deployed products that reliably assess 'adequacy of scope' or discover subtle internal-control weaknesses in production are narrow and still require heavy human validation. |
Examine the minutes of meetings of directors, stockholders, and committees to investigate the specific authority extended at various levels of management.
32CI 25–40 · exposure 33 · augmentation 63 · importance 4.0/5 · click for rater detail
Examine the minutes of meetings of directors, stockholders, and committees to investigate the specific authority extended at various levels of management.
32| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services show moderate AI adoption, but this specific compliance task remains heavily human-driven due to regulatory requirements and the need for defensible human accountability in examination records. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show fast AI adoption for document review and compliance support, but examiner roles within regulatory bodies tend to adopt more cautiously due to audit and legal accountability requirements. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist by pre-processing minutes, flagging authority keywords, and organizing findings for review, meaningfully raising examiner productivity without removing the required human judgment on authority validation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can efficiently extract, summarize, and cross-index board minutes to highlight delegation-of-authority language, significantly speeding up the examiner's review while the examiner retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can extract and summarize meeting minutes using document processing and NLP, but determining management authority levels requires contextual understanding of organizational structure, legal conventions, and implicit delegation patterns that current systems struggle with reliably at scale. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can readily read and summarize meeting minutes and flag authority-related clauses, but verifying regulatory implications and forming an examiner's judgment on authority scope still requires human interpretation and cross-referencing with regulations. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial examiners are often licensed professionals (CPA, etc.) and regulators typically require human judgment and sign-off on governance findings; liability for missed authority gaps creates strong organizational and legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examination is a licensed, regulated function where findings carry legal and supervisory weight, requiring a credentialed examiner to sign off on authority determinations, creating strong institutional and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted document review reduces labor but requires expert human oversight to verify authority findings; the combined cost of AI processing plus mandatory human validation remains comparable to direct human review for this high-stakes task. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | LLM-based document review is cheap per page, but the need for expert oversight and verification against regulatory frameworks narrows the cost advantage to roughly comparable once quality assurance is included. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While document analysis tools exist, no deployed product reliably extracts authority chains and validates delegation across corporate governance documents with the precision required for regulatory compliance; most solutions require significant human interpretation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Document-analysis and summarization products exist and are used in compliance/legal workflows, but no deployed product reliably performs the specific regulatory judgment of validating delegated authority scope in financial exams at production scale. |
Prepare reports, exhibits, and other supporting schedules that detail an institution's safety and soundness, compliance with laws and regulations, and recommended solutions to questionable financial conditions.
29CI 23–36 · exposure 33 · augmentation 63 · importance 4.5/5 · click for rater detail
Prepare reports, exhibits, and other supporting schedules that detail an institution's safety and soundness, compliance with laws and regulations, and recommended solutions to questionable financial conditions.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Banking regulators and financial institutions operate in a highly regulated, risk-averse environment with slow digital transformation. Adoption of AI for core supervisory and compliance reporting remains minimal; organizations prefer human examiners for liability and legitimacy reasons. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show strong AI adoption, but regulatory examination functions are more conservative and cautious given compliance and liability concerns, so uptake is moderate. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist examiners by accelerating data aggregation, generating initial report templates, and flagging anomalies in financial metrics, meaningfully raising human productivity on the data and narrative synthesis portions while the examiner retains full responsibility for judgment and recommendations. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully speed up drafting, data extraction, and formatting of reports and schedules, letting examiners focus on judgment-based analysis and final recommendations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can extract data and generate drafts of routine sections, the task requires expert judgment on complex financial conditions, regulatory interpretation, and recommendation synthesis that cannot be reliably automated end-to-end. Current AI systems cannot consistently achieve the regulatory compliance rigor and context-dependent recommendations required for a 50% time-saving threshold at equal quality. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft report sections, compile schedules, and summarize findings from structured data, but synthesizing regulatory judgment and recommending solutions to complex financial conditions requires human expertise and accountability. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Federal financial regulators (OCC, FDIC, Federal Reserve) require examiners with specific licensing and authority to conduct examinations and sign off on findings and recommendations. Regulatory frameworks mandate human professional judgment and accountability, creating strong legal and compliance barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examiners often operate under regulatory mandates requiring licensed/certified professionals to sign off on findings, creating strong institutional and legal barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for financial data extraction and report templating have moderate deployment and maintenance costs, but still require substantial human oversight to verify analysis and recommendations, keeping total cost close to or above that of a skilled examiner producing the report from scratch. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can reduce drafting and data-compilation time, but the human review, verification, and sign-off required for regulatory reports keeps overall costs closer to human-driven processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs the full task of preparing regulatory reports with the required accuracy and compliance guardrails. Some vendors offer document assembly and data aggregation tools, but material gaps remain in complex analysis, regulatory nuance interpretation, and recommendation formulation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some fintech/regtech tools assist with data aggregation and drafting, but no deployed product reliably produces complete examiner reports with regulatory recommendations without extensive human oversight. |
Investigate activities of institutions to enforce laws and regulations and to ensure legality of transactions and operations or financial solvency.
28CI 28–28 · exposure 25 · augmentation 75 · importance 4.2/5 · click for rater detail
Investigate activities of institutions to enforce laws and regulations and to ensure legality of transactions and operations or financial solvency.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Financial institutions have deployed AI for transaction monitoring and fraud detection, but these remain assistive tools; regulatory bodies (OCC, Federal Reserve) are cautious about automating enforcement and solvency determinations. Adoption is steady but not rapid, with human examiners remaining the bottleneck. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services and regulatory sectors have moderate AI adoption for surveillance and fraud detection, but examination and enforcement functions remain slower to adopt full automation due to compliance and legal risk. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered document review, anomaly flagging, regulatory database search, and pattern visualization meaningfully assist examiners in sifting large transaction volumes and identifying non-obvious red flags. These augmentations can substantially raise examiner productivity while keeping human judgment and accountability central to enforcement decisions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially aids investigators by flagging anomalies, summarizing large document sets, and detecting patterns in transactions, meaningfully boosting productivity while humans retain decision authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data analysis, pattern detection, and document review to identify anomalies, the task requires substantive human judgment to interpret complex regulatory frameworks, assess institutional risk, and determine enforcement actions. End-to-end automation would require AI to make legally binding determinations about solvency and compliance violations, which remains beyond current reliable AI capabilities. |
| Task automatability | claude-sonnet-5 | 2/5 | Investigation requires judgment, following leads, interviewing personnel, and interpreting ambiguous evidence within a legal enforcement context, which AI cannot fully replicate end-to-end today. AI can accelerate data analysis subcomponents but cannot conduct the full investigation autonomously. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory frameworks (Dodd-Frank, OCC rules, FDIC oversight) vest authority to investigate and enforce compliance in licensed examiners and institutions; regulators mandate human judgment and accountability in enforcement actions. Liability for incorrect determinations and legal standing of enforcement actions create strong legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examination is a regulated function often requiring credentialed examiners with legal authority to compel records, testify, and make binding determinations, creating strong licensing and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools for document analysis and transaction screening can reduce some grunt work, but the expertise required of financial examiners (regulatory knowledge, judgment, accountability) commands high hourly rates. The cost of integration, oversight, and maintaining compliance systems approaches or exceeds the loaded wage for many tasks. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools reduce analyst hours on data review but still require expensive human oversight, legal judgment, and documentation, keeping overall cost comparable to or only modestly below human-only investigation costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Products exist for anomaly detection and transaction monitoring in financial institutions, but deployed systems are narrow in scope and typically flag suspicious activity for human review rather than conducting independent investigations. No mature products perform the full investigatory, interpretive, and enforcement determination at scale without significant human oversight. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist for transaction monitoring, anomaly detection, and document review, but no deployed system performs full regulatory investigations independently in production; human examiners remain central. |
Establish guidelines for procedures and policies that comply with new and revised regulations and direct their implementation.
26CI 25–28 · exposure 25 · augmentation 63 · importance 3.8/5 · click for rater detail
Establish guidelines for procedures and policies that comply with new and revised regulations and direct their implementation.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services adoption of AI for core regulatory compliance tasks remains cautious and heavily supervised; pilots exist but full automation of guideline establishment is rare due to regulatory conservatism, risk aversion, and the need for human sign-off in this highly regulated domain. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services is a fast-adopting sector for AI generally, but compliance policy-setting functions are more conservative and still mostly pilot-stage due to regulatory risk. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by analyzing new regulations, suggesting policy language, identifying gaps in existing procedures, and helping with documentation, enabling financial examiners to work more efficiently while they retain final judgment and accountability. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully accelerate regulatory research, drafting policy language, and identifying compliance gaps, substantially boosting examiner productivity while humans retain final judgment and direction. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can help draft policy language and identify regulatory requirements from text, establishing guidelines requires domain expertise, legal judgment, and organizational context that current systems cannot fully synthesize. The task demands understanding nuanced regulatory intent and organizational constraints, making end-to-end automation with 50% time savings unlikely. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires interpreting new regulations, exercising judgment on institutional risk, and directing organizational implementation—AI can draft summaries and gap analyses but cannot own the accountable policy-setting and directing function end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: regulatory authority frameworks often require licensed professionals to sign off on compliance procedures, legal liability for incorrect guidelines falls on the organization, and financial regulators typically mandate human accountability for policy establishment and implementation oversight. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examiners often operate under statutory authority and institutional accountability structures; policy-setting typically requires sign-off from licensed/authorized personnel and carries significant liability exposure. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The specialized expertise required for regulatory guideline establishment commands high human wages, and AI solutions still require expert oversight, legal review, and iterative refinement, making the all-in cost per task nearly as expensive as hiring a skilled financial examiner. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply summarize regulatory text, but the human oversight, legal review, and organizational direction needed keep overall cost comparable to or only modestly below the human-led process. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs the full scope of establishing and directing implementation of compliance guidelines. AI can assist with regulatory text analysis and draft generation, but no production system independently handles the strategic, legal, and organizational components required for this task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Regulatory-tech tools exist to track rule changes and flag compliance gaps, but no deployed product independently establishes and directs implementation of institutional policy at scale. |
Evaluate data processing applications for institutions under examination to develop recommendations for coordinating existing systems with examination procedures.
26CI 25–28 · exposure 25 · augmentation 63 · importance 3.5/5 · click for rater detail
Evaluate data processing applications for institutions under examination to develop recommendations for coordinating existing systems with examination procedures.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial regulatory institutions move slowly on AI adoption for core examination functions due to compliance requirements, risk aversion, and the need for human accountability. While data analytics tools are adopted, autonomous recommendation systems for examination coordination remain in pilot phases. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show moderate-to-fast AI adoption for data analysis, but specialized regulatory examination functions adopt more cautiously due to compliance and audit-trail requirements. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist examiners by summarizing data processing application features, flagging potential inconsistencies, and drafting analysis sections, meaningfully raising productivity while the human examiner retains control over final recommendations and regulatory judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist examiners by analyzing data processing systems, flagging anomalies, and drafting portions of recommendations, significantly speeding up the research and documentation phases while the examiner retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can analyze data processing applications and flagging inconsistencies, the task requires integrating domain-specific regulatory knowledge with institution-specific context to develop actionable recommendations. Current systems cannot reliably coordinate existing systems with examination procedures end-to-end without substantial human oversight and judgment. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires nuanced judgment about aligning an institution's specific IT systems with regulatory examination procedures, which involves contextual reasoning AI can partially support but not fully replace end-to-end today.dependencies on domain expertise and institution-specific context limit full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial examiners operate under strict regulatory frameworks (federal banking regulators) where recommendations must be defensible and often require sign-off by licensed professionals. Liability asymmetry for incorrect system recommendations is high, and regulators typically require human accountability for examination findings. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examination work is subject to regulatory frameworks requiring qualified, often licensed examiners to sign off on findings and recommendations, creating significant institutional and legal barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The setup, validation, and ongoing oversight required to use AI for regulatory evaluation makes the cost comparable to or higher than a human examiner, especially when accounting for liability and correction costs if recommendations fail regulatory scrutiny. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Given the specialized judgment and regulatory context required, AI tools mainly assist rather than replace the examiner, so full-task AI cost including necessary human oversight is not dramatically cheaper than a skilled examiner's time. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Limited products exist that can evaluate data processing applications in the financial regulatory context. Existing AI systems can draft analyses but lack reliable deployment in production financial examination environments where error costs are high and regulatory compliance is mandatory. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | No deployed products autonomously evaluate data processing applications and generate examination-coordination recommendations in production; this remains largely a manual analyst function with AI used only for narrow sub-tasks like data extraction. |
Recommend actions to ensure compliance with laws and regulations, or to protect solvency of institutions.
24CI 20–28 · exposure 25 · augmentation 75 · importance 4.7/5 · click for rater detail
Recommend actions to ensure compliance with laws and regulations, or to protect solvency of institutions.
24| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial institutions deploy AI for data processing and pattern detection in compliance, but AI-generated recommendations for regulatory actions remain rare in production. Risk-averse regulatory culture and slow decision cycles in banking limit velocity; pilots and vendor POCs are common, but displacement of examiner judgment has not materialized at scale. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show fast AI adoption for analytics and monitoring, but regulatory examination functions remain cautious, with pilots more common than full production deployment for judgment-based recommendations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered compliance monitoring, automated document review, and regulatory-change alerting meaningfully assist financial examiners by reducing time spent on data gathering and flagging anomalies. Examiners retain judgment on remediation strategy, but AI augmentation can raise their productivity and coverage by 30–50% on screening and diagnostic phases. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly assist by flagging risk indicators, summarizing regulations, and drafting reports, substantially boosting examiner productivity while the human retains decision authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can analyze regulatory requirements and flag compliance gaps, but recommending actions to ensure compliance requires contextual judgment about institutional risk tolerance, legal strategy, and remediation trade-offs that exceed current AI capability. The task demands synthesis of complex regulatory interpretation with institutional-specific factors that AI struggles to handle reliably without human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires professional judgment, synthesis of regulatory context, and institutional risk assessment that current AI cannot reliably produce end-to-end; AI can draft supporting analysis but not independently generate actionable compliance recommendations with accountability.ractéristiques.You |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial examiners operate under strict regulatory authorization; regulators (OCC, FDIC, Fed) oversee examination practices and require licensed examiners to sign off on institutional recommendations. Liability asymmetry is acute: a flawed AI recommendation that misses a solvency risk creates institutional and reputational harm, locking in human accountability. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Financial examiners are licensed/credentialed professionals operating under strict regulatory mandates, and their recommendations often carry legal and institutional accountability that requires a qualified human to sign off. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI compliance-scanning tools cost tens to hundreds of thousands annually and require expert oversight to vet recommendations. A financial examiner's loaded cost is roughly $100–150k per year; the all-in AI cost (software, integration, human review) is comparable or higher given the verification burden. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate draft analysis, but the human oversight, verification, and liability review needed for regulatory recommendations keep all-in costs comparable to or only modestly below human examiner costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI tools exist to scan regulations and identify compliance issues, no deployed product reliably generates actionable compliance recommendations that institutions can rely on without substantial human legal review. Regulatory compliance is high-error-cost; vendors lack production evidence of replacing human examiner judgment. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some compliance-support and risk-analytics tools exist, but no deployed product autonomously issues authoritative regulatory or solvency recommendations in production at scale. |
Verify and inspect cash reserves, assigned collateral, and bank-owned securities to check internal control procedures.
24CI 20–28 · exposure 25 · augmentation 63 · importance 3.2/5 · click for rater detail
Verify and inspect cash reserves, assigned collateral, and bank-owned securities to check internal control procedures.
24| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Banking and financial regulation remain heavily governed by compliance and audit requirements; adoption of AI for core examination functions has been cautious and limited, with most deployed systems assisting rather than replacing human examiners. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Financial services broadly show fast AI adoption for analytics and fraud detection, but the specific regulated examination/audit function lags with mostly pilot-stage tools. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can significantly augment productivity by automating data aggregation, flagging inconsistencies, and pre-screening securities and collateral details, allowing examiners to focus judgment on complex exceptions and control assessment decisions. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist by flagging anomalies in transaction data, cross-referencing records, and speeding up data aggregation, but the core verification and judgment steps remain human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Parts of the task (data collection, basic reconciliation checks against records) can be automated, but verifying collateral quality and assessing control procedures requires subjective judgment and on-site inspection that current AI cannot reliably do end-to-end at the quality standard required for regulatory compliance. |
| Task automatability | claude-sonnet-5 | 2/5 | The physical/procedural verification of cash reserves, collateral, and securities requires on-site inspection, document authentication, and judgment calls that current AI cannot fully replicate end-to-end, though data reconciliation portions could be assisted. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Regulatory frameworks (OCC, Federal Reserve, FDIC) require licensed bank examiners to perform or directly supervise cash and collateral verification; liability, audit trail, and the need for legally defensible human certification create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Financial examination is a regulated function often requiring licensed examiners to certify findings, with significant liability if verification is inaccurate, creating strong barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI can handle data reconciliation cheaply, but the high-stakes nature of bank examination means human oversight and final sign-off remain mandatory, making the total cost of the AI-assisted process comparable to or higher than the baseline human examiner cost. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can cheaply process large data reconciliations, but the human oversight, physical verification, and liability-bearing sign-off still require examiner time, keeping blended costs closer to human-level. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Limited deployed products exist for this specific task; while AI can extract and cross-check numerical data from financial systems, assessing the adequacy of internal control procedures and physical collateral inspection remain largely human-centric and not reliably automated in production at scale in the regulatory context. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some fintech and audit tools can flag discrepancies in transaction records, but no deployed product independently performs full verification/inspection of collateral and securities holdings reliably in production. |
Train other examiners in the financial examination process.
23CI 16–30 · exposure 17 · augmentation 63 · importance 3.8/5 · click for rater detail
Train other examiners in the financial examination process.
23| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial regulation moves conservatively; training practices remain largely human-centered and in-person/synchronous. While some regulatory bodies pilot e-learning, autonomous AI-led training adoption is minimal in this highly oversight-conscious sector. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Financial regulatory and examination sectors are cautious adopters of AI for training functions, with pilots emerging but production-scale AI-led training uncommon. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist trainers by drafting lesson materials, generating scenarios, and providing automated quizzes or reference summaries, usefully reducing preparation burden. However, the core mentoring and judgment-intensive assessment remain human-centric. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist trainers by generating case studies, quizzes, explanatory content, and simulated scenarios, significantly enhancing training efficiency while humans remain in charge. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Training requires nuanced judgment about learner needs, adaptive pedagogical sequencing, and real-time coaching—tasks that demand deep contextual understanding and human mentorship. Current AI cannot reliably replicate the personalized guidance, credibility, and accountability inherent in professional training. |
| Task automatability | claude-sonnet-5 | 2/5 | Training involves live instruction, mentoring, and adapting to trainee needs and organizational context, which current AI cannot fully replicate end-to-end, though it can help create materials.ateway.rating.2 |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Banking regulators and examination authorities typically require training be delivered or certified by qualified, credentialed examiners; there are implicit professional-practice norms and likely regulatory expectations that human expertise oversee competency certification in this safety-critical domain. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for the trainer role itself, but regulatory expectations for examiner competency and institutional preference for experienced human mentors create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-generated training content can reduce some preparation costs, but delivering high-fidelity examination training still requires experienced human trainers. The savings from automation do not yet offset the need for qualified examiner-instructors to validate and deliver training. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply produce supplementary training materials, but human trainers still needed for mentoring, case discussion, and certification sign-off, keeping overall cost comparable to human-led training. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate training materials and answer procedural questions, no deployed system reliably conducts end-to-end examination training with the mastery-based assessment and professional credibility that regulators and trainees expect. Products exist for support but not autonomous delivery. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist for generating training content and quizzes, but no deployed system autonomously trains examiners in the nuanced judgment-based financial examination process at scale. |
Review applications for mergers, acquisitions, establishment of new institutions, acceptance in Federal Reserve System, or registration of securities sales to determine their public interest value and conformance to regulations, and recommend acceptance or rejection.
23CI 20–25 · exposure 25 · augmentation 63 · importance 3.5/5 · click for rater detail
Review applications for mergers, acquisitions, establishment of new institutions, acceptance in Federal Reserve System, or registration of securities sales to determine their public interest value and conformance to regulations, and recommend acceptance or rejection.
23| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services and banking are highly regulated; adoption of AI for high-stakes approval decisions remains cautious and pilot-stage despite digitization. Compliance automation is more mature, but autonomous decision-making on merger/acquisition public-interest determinations is rare in production, constrained by regulatory mandate and institutional risk aversion. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government financial regulatory agencies are typically slower and more cautious adopters of AI for high-stakes decisions compared to private-sector finance, though some analytics tools are piloted. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by automating document ingestion, flagging regulatory inconsistencies, summarizing financial metrics, and compiling compliance checklists—raising examiner productivity in research and preliminary screening phases. However, the core judgment task (public interest assessment and recommendation) remains human-driven, limiting augmentation to intermediate workflow support rather than transformative productivity gain. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully speed up document review, cross-referencing regulations, and flagging anomalies or risks, letting examiners focus on judgment calls and final recommendations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist in document parsing, regulatory checklist verification, and preliminary compliance screening, the task fundamentally requires judgment about 'public interest value'—a nuanced legal and policy determination that depends on context, precedent interpretation, and discretionary reasoning. Current AI cannot reliably perform the full end-to-end task of synthesizing complex merger documents, assessing market impact, and making binding recommendations at ≥50% time savings without substantial human oversight and correction. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with document review and flagging regulatory issues but the core judgment—weighing public interest, systemic risk, and regulatory conformance for approval/rejection—requires accountable human decision-making that current systems cannot fully replace. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Hard barriers exist: federal financial regulations (12 U.S.C., banking statutes) require specific licensed examiners to review and certify findings; liability for approval decisions rests with the institution and regulator; and the task involves legal judgment on securities registration and Fed membership eligibility that cannot be delegated to unlicensed systems. Human sign-off is legally mandated. |
| Adoption barriers | claude-sonnet-5 | 5/5 | This is a statutorily assigned government/regulatory function requiring a credentialed examiner's judgment and legal accountability; formal decisions must be made or approved by authorized personnel. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Integration and oversight costs are substantial because human financial examiners must validate all recommendations, understand liability exposure, and retain decision authority. AI tools reduce document review overhead but do not eliminate the need for expert human review, making the combined cost comparable to or potentially higher than traditional human review without AI assistance. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI could cheaply pre-screen documents, but the overall task still requires expensive expert examiner time for judgment and sign-off, keeping blended costs closer to human-level rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production financial regulatory system currently deploys AI to autonomously review merger/acquisition applications and make acceptance/rejection recommendations. Regulatory compliance products exist for document classification and checklist verification, but they operate at narrow scope with meaningful error rates and require extensive human validation before submission to regulators. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some regtech and document-analysis tools are deployed for compliance screening, but no production system independently evaluates and recommends acceptance/rejection of merger or securities applications at regulatory agencies. |
Resolve problems concerning the overall financial integrity of banking institutions including loan investment portfolios, capital, earnings, and specific or large troubled accounts.
20CI 15–25 · exposure 20 · augmentation 63 · importance 4.2/5 · click for rater detail
Resolve problems concerning the overall financial integrity of banking institutions including loan investment portfolios, capital, earnings, and specific or large troubled accounts.
20| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Banking institutions have adopted AI for narrow tasks (fraud detection, compliance monitoring) but adoption of AI-driven resolution of systemic financial integrity problems remains limited; heavy regulatory scrutiny, risk aversion, and requirement for human accountability slow velocity in this sensitive domain. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Banking regulation is a conservative, highly regulated sector where AI adoption for core supervisory judgment remains in early pilot stages rather than production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can meaningfully assist by surfacing portfolio anomalies, stress-testing scenarios, and flagging problematic accounts, thereby improving an examiner's productivity and decision quality; however, final diagnosis and strategic resolution remain heavily human-dependent. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly assist examiners by flagging anomalies, summarizing loan portfolios, and modeling risk scenarios, meaingfully speeding up the analytical groundwork even though final resolution remains human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data analysis, anomaly detection, and portfolio assessment, resolving financial integrity problems requires complex judgment calls on risk mitigation, regulatory interpretation, and institution-specific strategy that AI cannot fully replicate. Only specific sub-components (like flagging suspicious transactions) approach the 50% threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | This task requires complex judgment, negotiation, regulatory interpretation, and accountability for systemic risk decisions that current AI cannot reliably perform end-to-end.i AI can assist with data analysis but cannot resolve the underlying institutional problems. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong regulatory and legal barriers apply: financial examiners operate under banking regulations (OCC, FDIC, Fed mandates), and resolution of institution-level financial integrity issues typically requires licensed, accountable human judgment and sign-off, with liability and compliance risk making full automation infeasible. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Financial examination is a licensed, regulated function often requiring statutory authority, sign-off, and accountability that legally must rest with a qualified human examiner or regulator. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | High-skilled financial examiners command significant salaries, and AI oversight, integration, and validation costs remain material for this sensitive domain; current AI inference costs do not yet achieve an order of magnitude advantage when factoring in necessary human review and accountability. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply process financial data, the human oversight, regulatory judgment, and liability required for actual problem resolution keep all-in costs comparable to or higher than pure automation savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs end-to-end resolution of banking institution financial integrity issues; existing systems handle narrower tasks like fraud detection or risk scoring but do not systematically resolve portfolio and capital problems at scale in production banking environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously resolves bank financial integrity problems; existing tools are analytical aids used by human examiners, not decision-making replacements. |
Plan, supervise, and review work of assigned subordinates.
12CI 7–16 · exposure 8 · augmentation 50 · importance 4.1/5 · click for rater detail
Plan, supervise, and review work of assigned subordinates.
12| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Financial services and regulated industries adopt AI cautiously for supervisory tasks due to compliance risk. Most firms use AI for analytics dashboards or scheduling assistance, not autonomous team management; adoption remains experimental and narrow. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While financial services broadly adopts AI tools, actual delegation of supervisory/managerial authority to AI is essentially nonexistent in practice. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist managers by flagging performance anomalies, summarizing team metrics, suggesting coaching points, or automating scheduling—practical augmentations that raise manager productivity without removing human judgment from the core supervisory relationship. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist by tracking task status, flagging errors in subordinates' work, and summarizing performance data, but the core planning and supervisory judgment remains human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Managing subordinates requires real-time interpersonal judgment, conflict resolution, and adaptive coaching tailored to individuals—tasks where current AI lacks agency and human trust. While AI could assist with scheduling or performance data aggregation, end-to-end supervision (assigning work, reviewing quality, coaching, discipline) remains fundamentally human-dependent. |
| Task automatability | claude-sonnet-5 | 1/5 | Supervising and reviewing subordinates' work requires personnel management, mentoring, and interpersonal judgment that current AI cannot perform end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Labor law, employment contracts, fiduciary duty, and regulatory frameworks (especially in financial services) require a human manager to be responsible for hiring, performance evaluation, and termination decisions. Liability and legal accountability create hard barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Supervisory authority typically requires organizational accountability, HR/legal responsibility, and managerial judgment that cannot be delegated to software without a human in the authority role. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Autonomous AI supervision would require human oversight anyway; the cost of deploying an AI system plus a human manager to audit it exceeds simply retaining the manager. AI tools may reduce administrative time, but cannot replace the role at lower total cost. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this managerial function, so cost comparison favors the human by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No production system today reliably supervises human staff autonomously. AI dashboards and alerts exist, but the core act of planning work, reviewing performance, and managing people remains a human responsibility with legal and organizational accountability. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product manages human subordinates or performs supervisory planning; this remains a research-stage aspiration at best. |
Direct and participate in formal and informal meetings with bank directors, trustees, senior management, counsels, outside accountants, and consultants to gather information and discuss findings.
1CI 0–3 · exposure 0 · augmentation 38 · importance 4.7/5 · click for rater detail
Direct and participate in formal and informal meetings with bank directors, trustees, senior management, counsels, outside accountants, and consultants to gather information and discuss findings.
1| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Banking regulation and financial examination are highly regulated domains with strong institutional resistance to removing human examiners from direct stakeholder engagement. Adoption velocity is negligible. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Banking regulatory examination is a conservative, compliance-heavy sector where AI adoption for core supervisory interactions remains slow and largely confined to back-office support. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with preparation (document analysis, summary generation, agenda drafting) or post-meeting analysis, but the core meeting conduct itself leaves little room for in-the-loop augmentation given the authority and personal accountability required. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help prepare briefing materials, summarize prior findings, and draft meeting agendas or notes, but the core meeting conduct itself is not significantly transformed. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task fundamentally requires human judgment in interpreting responses, building trust, and navigating complex interpersonal dynamics with senior stakeholders. Current AI cannot autonomously conduct meaningful meetings, read room dynamics, or make real-time contextual decisions about what information to pursue in conversation. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires live human interpersonal interaction, negotiation, and real-time judgment in high-stakes regulatory meetings, which current AI cannot conduct end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | This task has hard regulatory and legal barriers: financial examiners are federally appointed positions requiring specific licensing and authority, and stakeholders have a legal right to meet with authorized human representatives. Liability and regulatory requirements mandate human sign-off and presence. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Financial examiners are government-authorized officials whose formal meetings and findings carry legal and regulatory weight, requiring licensed human judgment and accountability. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying an AI system that could substitute for a financial examiner in stakeholder meetings (if it were even possible) would far exceed the loaded wage of an experienced examiner, and no such system exists at scale. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this interactive, authority-laden function, so cost comparison favors the human examiner entirely. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can reliably conduct independent meetings with C-suite executives and external parties to gather sensitive financial information and discuss examination findings. This requires human presence, authority, and accountability. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product directs or participates in formal regulatory meetings with bank executives and counsels; this remains firmly a human-led activity. |
Confer with officials of real estate, securities, or financial institution industries to exchange views and discuss issues or pending cases.
1CI 0–3 · exposure 0 · augmentation 38 · importance 2.9/5 · click for rater detail
Confer with officials of real estate, securities, or financial institution industries to exchange views and discuss issues or pending cases.
1| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Financial services regulation is heavily rule-bound and risk-averse. Agencies have not moved to automate official examiner-to-stakeholder conferences; human-led dialogue remains the standard and required practice in regulatory environments. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While financial services broadly adopt AI tools, the specific interpersonal, authority-based conferring function sees negligible AI penetration or displacement. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with pre-conference research, briefing document generation, or post-conference memo drafting, but the core task—live conferencing with officials—remains essentially human-led. Augmentation is marginal. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help examiners prepare talking points, summarize case files, or draft follow-up correspondence, moderately aiding preparation without touching the actual conferring itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task is fundamentally interpersonal negotiation and relationship-building requiring real-time dialogue, contextual judgment, and professional authority. Current AI cannot conduct substantive, two-way conferencing with external stakeholders where nuance, trust, and accountability matter. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a live interpersonal negotiation and information-exchange activity requiring real-time judgment, relationship management, and authority representation that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Financial examiners operate under regulatory authority and must personally represent their agency in official discussions. Liability, legal standing, and regulatory requirements mandate that a licensed examiner conduct or sign off on all material conferencing with external parties. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Financial examiners hold statutory authority and accountability; representing an agency or institution in discussions of pending cases requires a credentialed human with legal standing and liability exposure. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task requires human presence and authority; AI tools that might assist (drafting, research) cost extra on top of the examiner's time rather than replacing it. Total cost remains dominated by the examiner's loaded wage. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this interactive, authority-laden conferring task, so cost comparison favors the human by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably conducts independent conferences with industry officials on pending regulatory or legal cases. AI might draft talking points or summarize positions, but cannot substitute for the human examiner's participation and decision-making. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts substantive regulatory conferences or negotiations with industry officials on behalf of an examiner; this remains firmly human-only in practice. |
Related occupations — Business & Financial Operations
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.