New Accounts Clerks
43-4141.00Median wage $47,670/yr36,860 employed (US)Rank #44 of 923 scored · top 5% by substitution
Interview persons desiring to open accounts in financial institutions. Explain account services available to prospective customers and assist them in preparing applications.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution59
Exposure58
Augmentation64
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
15 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
27%
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%58
panel mean rating 3.3/5 → substitution pressure 58/100
Technical feasibility todayw 20%57
panel mean rating 3.3/5 → substitution pressure 57/100
Cost vs. human wagew 15%69
panel mean rating 3.8/5 → substitution pressure 69/100
Adoption barriersw 20%inverted — strong barriers lower the score51
panel mean rating 2.9/5 (barrier strength) → substitution pressure 51/100
Sector adoption velocityw 10%63
panel mean rating 3.5/5 → substitution pressure 63/100
Task breakdown (15 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.
Duplicate records for distribution to branch offices.
97CI 97–97 · exposure 100 · augmentation 38 · importance 2.7/5 · click for rater detail
Duplicate records for distribution to branch offices.
97| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services, insurance, and administrative organizations have been automating data replication and record distribution for decades; modern RPA and cloud-based solutions accelerate further adoption, particularly in larger, digitized firms. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Banking and financial services are among the more digitized industries with strong existing adoption of document automation and workflow systems for routine back-office tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Once automated, this task offers minimal opportunity for human-AI collaboration; the clerk may verify or manage the process, but the duplication itself has no meaningful augmentation mode. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Where duplication still involves human oversight or exception handling, automation tools speed the process, though the task is simple enough that augmentation offers only moderate additional value beyond full automation. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Duplicating records for distribution is a purely mechanical, rule-based task that modern RPA and data management systems can perform end-to-end with high speed and perfect accuracy, easily achieving >50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 5/5 | Duplicating and distributing records is a purely mechanical, digital data-handling task that off-the-shelf automation (document management systems, scripts, RPA) can fully perform with major time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | No licensing, regulatory sign-off, or human-contact requirement constrains automating record duplication; most organizations face no legal or policy barrier to switching to automated systems. |
| Adoption barriers | claude-sonnet-5 | 1/5 | This is a purely administrative, non-regulated clerical action with no licensing or human sign-off requirement standing in the way of automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated record duplication costs pennies per transaction in infrastructure and maintenance, whereas a clerk's fully-loaded wage (salary, benefits, facilities) is $30–50k annually for relatively few duplication tasks, making AI orders of magnitude cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated file copying/distribution costs a tiny fraction of a cent per transaction versus paying a clerk's wage for manual duplication and mailing/filing. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Record duplication and distribution is handled reliably by deployed systems across finance, banking, and administrative sectors today; document management and database replication tools execute this at scale in production environments. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Mature, widely deployed enterprise software (document management, RPA, cloud sync tools) already handles record duplication and multi-branch distribution reliably at scale in production banking environments. |
Collect and record customer deposits and fees and issue receipts, using computers.
87CI 79–95 · exposure 87 · augmentation 63 · importance 4.5/5 · click for rater detail
Collect and record customer deposits and fees and issue receipts, using computers.
87| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Financial institutions have already deeply adopted automated transaction processing, deposit recording, and digital receipts; this is standard practice, not emerging pilot work. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Banking and financial services are among the fastest and deepest adopters of digital/automated transaction processing, with self-service and automated systems already dominant in many markets. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems and banking software actively assist or fully replace human clerks on this task by auto-populating records, flagging discrepancies, and generating compliant receipts, measurably raising throughput and accuracy. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled systems assist clerks by auto-filling forms, flagging errors, and speeding reconciliation, though much of the task is already fully automated rather than merely augmented. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Collection and recording of deposits/fees with receipt issuance are entirely digital, rule-based transaction processes that current AI and banking APIs can execute end-to-end with full automation and significant time savings—no subjective judgment needed. |
| Task automatability | claude-sonnet-5 | 4/5 | Recording deposits, fees, and generating receipts is a structured, rules-based data entry task that current banking software and RPA/AI systems can largely handle end-to-end with substantial time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While banking is regulated and requires audit trails, the actual transaction recording is routinely automated in compliance frameworks; AML/KYC checks add some friction but do not legally require human clerks to perform the core deposit-and-receipt task. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Some regulatory recordkeeping and KYC compliance requirements exist, but the actual recording/receipting function itself is not legally required to be performed by a licensed human. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated transaction processing and receipt issuance cost pennies per transaction and scale to thousands daily, orders of magnitude cheaper than human clerk labor even when accounting for system maintenance and oversight. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated transaction processing systems cost fractions of a cent per transaction compared to a loaded clerk wage for the same volume of work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Banking and accounting software systems already automate deposit processing, fee recording, and receipt generation in production at scale across thousands of financial institutions worldwide. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Core banking platforms, ATMs, online banking, and RPA already perform deposit recording and receipt issuance reliably at scale in production, though some edge cases still route to staff. |
Obtain credit records from reporting agencies.
82CI 74–90 · exposure 87 · augmentation 75 · importance 4.4/5 · click for rater detail
Obtain credit records from reporting agencies.
82| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services and lending sectors have broadly digitized and deployed integrated account-opening platforms that automate credit retrieval. Adoption is mature in production banking and fintech environments, though smaller institutions may lag. |
| Sector adoption velocity | claude-sonnet-5 | 5/5 | Banking and financial services have deeply digitized credit checks for decades, with automated bureau integrations being standard practice industry-wide. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems can assist clerks by automatically fetching and formatting credit records, flagging discrepancies, and populating forms, substantially raising productivity while the human retains final review and decision authority over account approval. |
| Augmentation potential | claude-sonnet-5 | 4/5 | Where manual steps remain, automated retrieval tools significantly speed up clerks' work by eliminating manual lookup and re-entry tasks. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably retrieve credit records via APIs and automated workflows when given account holder information. The task is largely mechanical (sending requests, parsing responses) with minimal judgment, allowing >50% time savings at equal quality through end-to-end automation. |
| Task automatability | claude-sonnet-5 | 5/5 | Pulling credit records from bureaus (Equifax, Experian, TransUnion) via API is a well-defined, structured data retrieval task easily automated end-to-end with existing integrations. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While no formal license is strictly required to pull credit reports, regulatory requirements (FCRA compliance, fair lending rules, authorization from the applicant) create compliance and oversight friction that typically requires human sign-off or verification, limiting full substitution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | FCRA compliance requires permissible purpose documentation and consent, adding regulatory oversight, though the retrieval itself is not restricted to licensed humans. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated API calls cost pennies per request with minimal human oversight needed, compared to the loaded hourly wage of a clerk performing this repetitive retrieval task. The cost ratio is at least an order of magnitude in AI's favor. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | API-based automated credit pulls cost cents per transaction versus manual clerk time, an order-of-magnitude cost reduction. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple financial software platforms and accounting systems integrate directly with credit reporting agencies' APIs (Equifax, Experian, TransUnion) with proven production deployments. Error rates are low for standard requests, though edge cases may require human intervention. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Financial institutions already use production systems and automated pulls integrated into loan origination and account-opening software at scale today. |
Compile information about new accounts, enter account information into computers, and file related forms or other documents.
79CI 62–95 · exposure 83 · augmentation 75 · importance 4.6/5 · click for rater detail
Compile information about new accounts, enter account information into computers, and file related forms or other documents.
79| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Financial services and banking—sectors where this task is concentrated—are among the fastest adopters of RPA and AI-driven automation, with thousands of institutions already deploying these solutions in production. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Banking and financial services are moderately fast adopters of automation, with many institutions piloting or partially deploying account-opening automation, though full displacement is uneven across smaller institutions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | Even where human clerks remain, AI tools assist significantly by auto-populating fields, validating data quality, flagging errors, and automatically filing documents, raising clerk productivity substantially while they perform exception handling and verification. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI form-reading and data-entry tools significantly speed up compiling and entering account information, letting clerks focus on verification and exceptions rather than manual keying. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | This task is almost entirely routine data entry and document processing—compiling information, entering it into systems, and filing. Current RPA, OCR, and document-processing AI systems can handle all these components end-to-end with well over 50% time savings at equal or better accuracy than human entry. |
| Task automatability | claude-sonnet-5 | 4/5 | This is a structured data-entry and document-compilation workflow that AI-driven OCR/form-processing and RPA systems can largely automate, meeting the 50% time-saving bar for most routine cases. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While some financial institutions maintain compliance oversight and manual review steps, there are no legal requirements that a human must personally perform this data-entry task, and automation is already widely adopted. Minor friction exists around integration with legacy systems and regulatory audit trails, but these are surmountable. |
| Adoption barriers | claude-sonnet-5 | 3/5 | KYC/AML and identity verification regulations impose compliance and audit requirements, and institutions often keep human sign-off for new account approval, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | RPA and OCR infrastructure cost pennies per account after initial setup, whereas a clerk's fully loaded cost is $30–50k+ annually. The cost ratio is orders of magnitude in AI's favor for high-volume processing. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated data capture and entry systems cost a small fraction per transaction compared to a clerk's loaded wage, though integration and exception-handling oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, deployed products in banking, insurance, and financial services already automate account opening, data entry, and document filing at scale. Robotic process automation and intelligent document processing solutions are production-standard in these sectors. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Intelligent document processing and account-opening automation products exist in banking/fintech, but exception handling, KYC verification, and messy source documents still require human review, limiting fully reliable deployment. |
Answer customers' questions and explain available services, such as deposit accounts, bonds, and securities.
70CI 61–79 · exposure 67 · augmentation 75 · importance 4.5/5 · click for rater detail
Answer customers' questions and explain available services, such as deposit accounts, bonds, and securities.
70| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | The financial services sector is among the earliest adopters of AI chatbots and customer service automation, with major banks deploying these systems in production already, indicating rapid and deep adoption momentum. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Banking and financial services are among the faster adopters of AI customer service tools, with chatbots and virtual assistants already deployed at scale by major institutions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems effectively augment account clerks by providing instant access to service information, drafting responses, and surfacing relevant account details, allowing clerks to handle more complex inquiries and improve response quality and speed. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools substantially speed up clerks' ability to look up product details and draft explanations, letting them serve more customers with better consistency while retaining oversight for complex cases. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI chatbots and virtual assistants can handle many routine customer inquiries about standard deposit accounts and basic services, but complex questions involving personalized financial advice, regulatory compliance, or unusual account structures typically require human judgment and customization. |
| Task automatability | claude-sonnet-5 | 4/5 | Explaining standard account, bond, and deposit products is highly scriptable knowledge-Q&A work that current LLM-based chatbots handle well, though edge cases and account-specific actions still need human handling.dle handoff. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Banks face regulatory scrutiny on customer communications and must ensure accuracy in financial disclosures; many institutions prefer human oversight or hybrid models for liability reasons, though no strict licensing requirement prevents AI deployment. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement to explain basic deposit/bond products, but banks face some regulatory disclosure and suitability concerns plus customer preference for human reassurance on financial matters. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven customer service systems have very low marginal inference costs and can handle high volumes simultaneously, making them substantially cheaper than employing clerks to answer the same routine inquiries at scale. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | A conversational AI answering routine product questions costs a small fraction of a cent per interaction versus a loaded clerk wage, an order-of-magnitude or greater saving. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed chatbots and virtual customer service agents in banking are already performing this task in production at scale for straightforward questions about services, though they require human escalation for complex cases, demonstrating mature feasibility with some limitations. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Banks widely deploy chatbots and virtual assistants that answer product questions and explain services in production today, though accuracy and escalation-to-human gaps remain for complex queries. |
Perform foreign currency transactions and sell traveler's checks.
69CI 48–90 · exposure 70 · augmentation 50 · importance 2.8/5 · click for rater detail
Perform foreign currency transactions and sell traveler's checks.
69| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Financial services and fintech sectors have deeply and rapidly adopted automated currency exchange and traveler's check issuance; major banks and payment platforms now execute these transactions with minimal human intervention. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Retail banking and currency exchange services have been slow to fully automate this specific task, especially in physical branch settings, and traveler's checks have declined in relevance rather than being displaced by AI. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists clerks by flagging suspicious patterns, suggesting optimal exchange rates, and automating data entry, raising productivity on compliance and customer service aspects even when automation isn't complete. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-driven exchange rate calculators, fraud detection, and transaction logging assist clerks in speed and accuracy, though the core customer-facing and cash-handling elements remain manual. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Currency exchange calculations, verification of transaction details, and traveler's check issuance are highly structured, rule-based processes that current AI systems can fully automate. Integration with banking systems and compliance checks can be handled by existing fintech APIs and workflows, meeting the ≥50% time-saving threshold with no quality degradation. |
| Task automatability | claude-sonnet-5 | 3/5 | Currency conversion calculations and transaction processing can be automated, but the physical handling of cash/checks, ID verification, and customer interaction still require human presence in many contexts, limiting full end-to-end automation.rating below 4. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While regulatory and compliance oversight exists, there are few hard licensing barriers preventing automation; most regulatory frameworks focus on the organization's accountability rather than requiring a human to perform the transaction itself. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Currency exchange is subject to AML/KYC regulations and reporting requirements, and physical cash handling adds friction, though this is a lower barrier than tasks requiring licensed professional judgment. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated currency transaction processing is orders of magnitude cheaper than clerk labor—API calls and system overhead cost fractions of a cent per transaction versus hourly wages for manual handling. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated kiosks and digital forex platforms can be cheaper per transaction at scale, but compliance, cash reconciliation, and fraud monitoring still require oversight, keeping costs roughly comparable in many deployments. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed systems in digital banking platforms and currency exchange services routinely automate these tasks at scale; however, some organizations still maintain manual oversight for high-value or unusual transactions, preventing a perfect 5. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Self-service kiosks, ATMs, and online forex platforms exist and handle much of this reliably, but traveler's checks are largely obsolete and many currency exchange operations still rely on staffed counters for compliance and cash handling. |
Inform customers of procedures for applying for services, such as ATM cards, direct deposit of checks, and certificates of deposit.
66CI 52–79 · exposure 55 · augmentation 75 · importance 4.5/5 · click for rater detail
Inform customers of procedures for applying for services, such as ATM cards, direct deposit of checks, and certificates of deposit.
66| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Banking is a highly digitized sector with strong adoption of AI chatbots and self-service portals; many institutions already route initial account information through automated systems. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Retail banking has aggressively adopted chatbots, self-service portals, and virtual assistants for routine customer service and informational tasks over the past several years. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems effectively assist clerks by pre-drafting responses, retrieving policy details, suggesting applicable products, and handling routine inquiries, freeing staff to focus on complex or high-value customer needs. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up how clerks answer procedural questions via scripted assistants, knowledge bases, and real-time suggestion tools, improving consistency and speed. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI can draft informational content about banking procedures, but reliably informing customers requires understanding individual circumstances, addressing confusion in real-time, and adapting explanations—tasks that remain largely manual or heavily human-supervised today. |
| Task automatability | claude-sonnet-5 | 4/5 | This is a repetitive information-delivery task about standardized procedures that chatbots and conversational AI can handle end-to-end for most routine cases, though edge cases still need escalation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While financial institutions must comply with regulations (FCRA, Truth in Lending Act), the informational task itself is not legally restricted to licensed staff, and customers increasingly accept automated service delivery for routine procedures. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement to explain procedures, though banks maintain some human staff for compliance disclosures, trust-building, and handling customers who prefer or need in-person assistance. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-powered chatbots or self-service systems cost far less per interaction than a clerk's hourly wage, especially when handling multiple routine inquiries simultaneously at scale. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated conversational systems and FAQ bots cost a small fraction per interaction compared to a paid clerk's time for delivering standardized procedural information. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Chatbots and conversational AI exist in banking and can deliver standardized procedure information, but deployed systems often struggle with edge cases, regulatory nuance, and customer verification, requiring escalation to human staff. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Banks widely deploy chatbots, IVR systems, and virtual assistants that explain account opening and card/deposit procedures reliably today, though some complex or exception scenarios still route to humans. |
Execute wire transfers of funds.
60CI 51–69 · exposure 62 · augmentation 75 · importance 4.1/5 · click for rater detail
Execute wire transfers of funds.
60| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Financial services and accounting departments have been early adopters of AP automation and payment processing tools; Treasury Management Systems with API-driven workflows are now standard in mid-to-large enterprises. Adoption is widespread in digitized sectors, though smaller firms and community banks lag. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Financial services is a fast-adopting sector for transaction automation, with most major banks having already implemented automated wire processing and fraud detection systems at scale. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI dramatically assists clerks by auto-populating transaction details, validating routing numbers, flagging exceptions, and drafting transfer instructions, significantly reducing manual data entry and error-checking time. The human retains decision authority and approval responsibility, boosting their throughput. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-based fraud detection, verification, and workflow tools significantly speed up clerks' ability to process and validate wire transfers while keeping a human in the loop for authorization. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Wire transfers are highly structured transactions with clear inputs (account numbers, amounts, routing information) and deterministic outputs. Current AI systems can extract relevant data, validate formats, and initiate transfers through API integration with banking systems, though human authorization steps may remain for compliance. This achieves near 50% time saving at equal quality for the clerical portion. |
| Task automatability | claude-sonnet-5 | 3/5 | Wire transfer execution is a structured, rules-based data entry and verification task that automated banking systems can largely handle, but identity verification, fraud checks, and exception handling still require human oversight for full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Wire transfers are heavily regulated (AML, KYC, Patriot Act compliance); most jurisdictions and financial institutions legally require a licensed or authorized human to approve and sign off on transfers, even if an AI system populates the form. This creates a durable human-in-the-loop requirement. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Wire transfers are subject to strict banking regulations (BSA/AML, KYC, OFAC screening) requiring authorized personnel oversight and audit trails, and errors carry significant financial and legal liability. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven wire transfer automation costs only the marginal inference and API call fees (cents to dollars per transaction), vastly cheaper than the 15–30 minute human labor cost ($5–15 per transfer at median clerk wages). The cost ratio easily exceeds an order of magnitude in AI's favor. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated transfer processing systems handle high volumes at very low marginal cost per transaction compared to manual clerk processing, though compliance and oversight infrastructure adds some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple banking platforms and accounting software (e.g., Treasury Management Systems, AP automation tools) already perform wire transfer initiation and processing in production environments. However, most require human approval gates for fraud/compliance reasons, and integration complexity varies by institution, preventing a full 5 rating. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Banks widely deploy automated wire transfer processing and STP (straight-through processing) systems, but these operate within tightly controlled workflows with human review for exceptions, higher-value transfers, and fraud flags rather than fully autonomous execution. |
Process loan applications.
59CI 45–72 · exposure 62 · augmentation 75 · importance 4.6/5 · click for rater detail
Process loan applications.
59| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Financial services are among the earliest and fastest adopters of AI automation; major banks and lenders have deployed loan processing AI at scale for years, with measurable staff reduction and workflow acceleration widely reported in industry practice. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Banking and financial services are moderately fast adopters of AI for document processing and workflow automation, with many pilots and some production deployments, but comprehensive end-to-end AI loan processing remains uneven across institutions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists clerks by auto-populating fields, flagging inconsistencies, and pre-screening documents, reducing manual review time and error rates while the human retains oversight and judgment responsibilities—a clear productivity multiplier for human-in-the-loop operations. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up document verification, data entry, credit report parsing, and application triage, meaningfully boosting clerk productivity while humans retain oversight of final decisions and exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Loan application processing involves structured data extraction, verification against rules, and routing—tasks that current AI systems (document parsing, OCR, rule engines, workflow automation) can handle well. Human intervention remains needed for edge cases and final approval, but 50%+ time savings at equal quality is achievable with off-the-shelf tools. |
| Task automatability | claude-sonnet-5 | 3/5 | Data collection, document verification, and initial underwriting checks can be automated with existing systems, but exception handling, applicant communication, and final decisioning often still require human judgment, so only part of the end-to-end task meets the 50% time-saving bar without significant customization. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory requirements (Truth in Lending Act, Fair Credit Reporting Act, anti-discrimination rules) mandate clear audit trails and human sign-off on lending decisions; compliance oversight and liability asymmetry prevent full end-to-end automation and require retained human involvement in risk assessment and approval. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Lending is regulated (fair lending laws, KYC/AML, disclosure requirements) requiring documented compliance and often human review of certain decisions, but no specific individual license is generally required for a new accounts clerk to process routine applications. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven document processing and workflow automation cost significantly less than manual clerk labor per application when amortized across volume. A single platform serving thousands of applications drives per-unit cost well below loaded wages for clerical staff. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Loan processing software and AI tools reduce labor costs substantially for routine steps, but integration, compliance oversight, and exception handling keep total costs closer to parity with clerk wages rather than an order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (e.g., document automation platforms, loan origination software with AI modules) reliably perform parts of this workflow in production at many financial institutions. Error rates on structured data entry and basic eligibility checks are now acceptable, though manual review of complex cases remains standard practice. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Many banks deploy automated loan origination systems and AI-based document/credit checks in production, but these still require human review for edge cases, compliance sign-off, and exceptions, so reliability is narrow rather than fully autonomous. |
Perform teller duties as required.
57CI 45–70 · exposure 62 · augmentation 63 · importance 4.6/5 · click for rater detail
Perform teller duties as required.
57| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Major financial institutions have actively adopted AI and RPA in customer-facing and back-office teller functions over the past 5–10 years, with measured displacement of routine clerk roles. Adoption is deepest in larger banks and digital-first institutions. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Banking is a digitized, finance-sector industry with substantial fintech adoption, but branch teller automation has progressed unevenly and many institutions retain significant teller staffing. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists teller staff by automating routine data entry, compliance checks, and transaction initiation, allowing human clerks to focus on relationship-building, exception handling, and complex customer needs. This significantly raises clerk productivity in mixed human–AI workflows. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled systems (chatbots, ITMs, transaction software) assist tellers with routine inquiries and paperwork, improving throughput while humans manage exceptions and customer relations. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Routine teller duties—cash handling, account opening, transaction processing, customer verification—are highly structured and repetitive. AI-powered systems can automate a majority of these tasks (transaction recording, form filling, basic customer queries), achieving >50% time savings, though some physical handling and edge-case resolution may require human intervention. |
| Task automatability | claude-sonnet-5 | 3/5 | Routine teller functions like deposits, withdrawals, and balance inquiries can be handled by ATMs, ITMs, and digital banking, but cash handling and in-person exception processing still require human presence.4 |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Banking and financial services are heavily regulated; customer identification, fraud prevention, and account-opening workflows require compliance verification and often regulatory sign-off. These requirements create material friction, though they do not legally prohibit automation—they require embedded compliance controls. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but cash handling, fraud liability, KYC/AML compliance, and customer preference for face-to-face service in banking create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automated teller systems and RPA have very low marginal cost per transaction after deployment; inference and integration costs are negligible compared to loaded clerk wages. AI solutions are substantially cheaper per equivalent task completed. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | ITMs and digital banking infrastructure carry significant upfront capital and maintenance costs, offsetting labor savings, though at scale per-transaction cost can undercut a human teller. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Banking institutions have deployed AI and robotic process automation (RPA) in production for account opening, transaction processing, and verification workflows. These systems are operating at scale in major banks, though they typically handle routine cases and require human handoff for exceptions. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Interactive teller machines and mobile/online banking are deployed at scale in many banks, but branches still staff tellers for cash transactions, ID verification, and complex service issues, showing incomplete substitution. |
Refer customers to appropriate bank personnel to meet their financial needs.
47CI 36–59 · exposure 38 · augmentation 75 · importance 4.3/5 · click for rater detail
Refer customers to appropriate bank personnel to meet their financial needs.
47| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Banks have invested in chatbots and IVR systems for basic routing, but adoption remains mixed; many institutions still rely heavily on human staff for initial customer assessment, indicating moderate rather than deep adoption. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Retail banking has moderate AI adoption with chatbots and virtual assistants in production, but many institutions still rely on staff for nuanced referrals. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can effectively assist bank clerks by suggesting relevant departments or personnel based on customer profile data and common needs, significantly accelerating the matching process while the clerk maintains oversight and builds relationships. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can quickly suggest the right department or specialist based on customer data, significantly speeding up the clerk's referral decision while human judgment remains available. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Routing customers to appropriate personnel requires understanding individual financial needs and matching them to bank staff expertise—tasks involving judgment and customer assessment that current AI cannot reliably perform end-to-end without significant human supervision and setup. |
| Task automatability | claude-sonnet-5 | 3/5 | Routing/triage based on stated customer needs is a straightforward classification task that chatbots and IVR/agent systems can handle, though nuanced judgment about relationship fit may still need human input. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Banks may prefer human staff to build customer relationships and ensure compliance; customer expectations often favor direct human contact for financial decisions, creating organizational friction against pure automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for simple referrals, but banks often prefer human-guided routing for compliance and customer relationship reasons, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | An AI-assisted phone/chat system would approach parity with the human cost once development and oversight overhead are included, though labor cost varies significantly by geography and wage level. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated routing via chatbot/IVR is very cheap per interaction compared to a human clerk's time, though integration and oversight costs reduce the full 10x savings somewhat. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While chatbots can provide basic routing options, no deployed product reliably understands nuanced customer financial situations and makes accurate referrals at production quality without human intervention or review. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Bank chatbots and CRM-integrated routing tools exist and are used for basic inquiry triage, but complex referrals still commonly go through human staff, so scope is narrower than full automation. |
Investigate and correct errors upon customers' request, according to customer and bank records.
45CI 40–50 · exposure 45 · augmentation 75 · importance 4.1/5 · click for rater detail
Investigate and correct errors upon customers' request, according to customer and bank records.
45| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Banks have piloted AI for transaction monitoring and anomaly detection, but production deployment for autonomous error correction remains limited; most adoption sits at assisted review rather than full automation, reflecting regulatory caution and customer trust concerns. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Banking is a digitized, finance-sector environment with growing AI adoption in back-office operations, but customer-facing error resolution still largely relies on human clerks with AI as a support tool. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists clerks by surfacing discrepancies, organizing relevant records, and suggesting likely errors or corrective actions, allowing humans to validate and authorize fixes faster; this augmentation is already deployed in many banking environments and materially raises clerk productivity. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly speed up error detection, retrieve relevant records, and draft resolution steps, meaningfully boosting clerk productivity while humans remain responsible for final corrections. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can automate parts of error investigation (flagging discrepancies via record comparison, pulling relevant transaction history) but typically requires human judgment to understand context, customer disputes, and corrective action authorization. Full end-to-end automation with 50% time saving is achievable for routine, high-volume error types but not across the full scope of complex disputes. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can flag discrepancies and draft explanations by comparing records, but investigating root causes and authorizing corrections often requires judgment and system access that current off-the-shelf AI cannot fully replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Banking regulations (compliance, fraud investigation, customer authorization requirements) and liability asymmetry (errors cause reputational and financial harm) create strong friction; humans must often legally sign off on corrections, and customer contact/satisfaction requirements add human-contact mandates. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Bank record corrections often require verified identity, compliance checks, and audit trails, creating moderate regulatory and liability friction even though no strict licensing requirement exists. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI-driven error detection and investigation tools (inference + integration) approach parity with loaded clerk wages for high-volume routine cases, but the need for human oversight, remediation, and customer communication means the all-in cost remains roughly comparable to human performance. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply triage and surface likely errors, but human review, correction authority, and exception handling still add substantial cost, making overall savings moderate rather than dramatic. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Deployed systems exist for transaction monitoring, anomaly detection, and basic error flagging in banking (compliance software, AML platforms), but they operate with material false-positive rates and narrow scope, requiring human review and decision-making before correction is applied. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some banks deploy AI-assisted case management and anomaly detection tools, but reliable autonomous error investigation and correction in production remains limited and narrow in scope. |
Interview customers to obtain information needed for opening accounts or renting safe-deposit boxes.
41CI 25–56 · exposure 38 · augmentation 63 · importance 4.3/5 · click for rater detail
Interview customers to obtain information needed for opening accounts or renting safe-deposit boxes.
41| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Banks have deployed chatbots for initial triage and FAQs, but actual account-opening remains human-led in production; regulatory caution and liability concerns mean adoption of end-to-end AI remains in pilots rather than mainstream displacement. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Retail banking has rapidly adopted digital account opening and chat-based onboarding tools, reflecting fast adoption typical of financial services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist clerks by pre-filling forms, flagging missing fields, or suggesting follow-up questions based on customer profile, measurably speeding data entry and recall; however, the core interview judgment still rests with the human. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-assisted forms, pre-fill data, and chatbot triage significantly speed up the interview and data-collection process while a human remains available for verification and exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can conduct basic structured information collection via chatbots, the task requires real-time judgment about customer needs, sensitivity to context, and verification of identity—elements that introduce variability and risk. Current AI systems cannot reliably handle the full interview end-to-end with equal quality and meet the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 3/5 | Structured data collection via chatbots or voice agents can gather account information, but identity verification, compliance checks, and handling exceptions still often require human judgment or oversight, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Financial services regulation (KYC, AML, FCRA) typically requires that qualified humans verify identity, assess risk, and take responsibility for account decisions; liability for automated false positives in fraud detection or credit decisions creates strong legal and organizational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | KYC/AML regulations require identity verification and recordkeeping, and some institutions require human sign-off, creating moderate regulatory and liability friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI chatbot and oversight infrastructure costs are competitive with entry-level clerk wages, but the need for human review, exception handling, and compliance verification prevents an order-of-magnitude cost advantage. Integration and error-mitigation costs are significant. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated onboarding forms and conversational AI are substantially cheaper per interaction than a live clerk, though compliance-related oversight adds some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Chatbots and voice AI exist for basic intake, but deployed solutions are narrow, often requiring human fallback for edge cases, complex queries, or regulatory compliance verification. No mature production system reliably completes full account-opening interviews without substantial human oversight. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Banks deploy online account-opening flows and chatbots that collect customer info, but these are narrow-scope and often fall back to human staff for edge cases, KYC issues, or safe-deposit box arrangements. |
Schedule repairs for locks on safe-deposit boxes.
18CI 5–30 · exposure 8 · augmentation 38 · importance 3.2/5 · click for rater detail
Schedule repairs for locks on safe-deposit boxes.
18| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Banking operations are transitioning slowly to AI automation for back-office tasks; safe-deposit box scheduling is a low-volume, infrequent task with strong organizational preference for human accountability in secure environments. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Banking back-office clerical roles show slow, uneven AI adoption for such narrow physical-facility-adjacent tasks compared to core financial services automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide minor assistance by flagging overdue maintenance or suggesting repair vendors, but the task is fundamentally about human coordination with external parties and does not benefit substantially from AI augmentation. |
| Augmentation potential | claude-sonnet-5 | 3/5 | General AI scheduling and calendar tools can help a clerk draft repair requests, track vendor communications, and manage timelines, offering moderate assistance while the clerk retains oversight. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Scheduling repairs for physical locks requires coordination with external service providers, understanding safe-deposit box inventory, and managing facility access—tasks involving human judgment, external communications, and physical world knowledge that current AI cannot perform end-to-end without significant human intervention. |
| Task automatability | claude-sonnet-5 | 2/5 | This is a coordination/scheduling task requiring contact with vendors or maintenance staff and physical verification, which current AI can partially assist but not fully execute end-to-end reliably.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Banks are heavily regulated, and safe-deposit box maintenance involves security and fiduciary responsibilities; human oversight and sign-off by authorized personnel are typically required before external vendors access secure facilities. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for scheduling itself, but bank operational policies, vendor relationships, and physical security concerns around safe-deposit boxes create moderate organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The task requires minimal human time (likely 5–15 minutes per scheduling event) and involves low-wage clerical work; AI integration for such infrequent, specialized coordination would cost more than the human labor saved. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While generic scheduling assistants exist, integrating them for this niche, low-frequency task would cost more in setup than the marginal labor savings from a clerk who already handles it as one of many duties. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product reliably handles the full workflow of identifying faulty locks, contacting repair vendors, coordinating facility schedules, and confirming repairs in production banking systems today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product specifically handles safe-deposit lock repair scheduling; this is a narrow, low-volume administrative task with no dedicated automation solutions in production. |
Issue initial and replacement safe-deposit keys to customers, and admit customers to vaults.
4CI 0–7 · exposure 0 · augmentation 25 · importance 4.1/5 · click for rater detail
Issue initial and replacement safe-deposit keys to customers, and admit customers to vaults.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task is embedded in heavily regulated financial institutions with strict security protocols. Adoption of automation would require regulatory approval and fundamental redesign of physical banking infrastructure, both of which proceed at glacial pace. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While banking as a sector adopts AI in many back-office and digital functions, this specific physical vault-access task sees essentially no AI adoption or pilots given its physical security nature. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with administrative record-keeping or customer identity pre-verification, but cannot augment the core physical and security-critical aspects of key issuance and vault access management. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could assist with identity verification, scheduling, or record-keeping around key issuance, but it offers minimal direct augmentation to the physical act of handing over keys and admitting someone to a vault. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical interaction (handing keys, admitting customers to physical vaults) and real-time identity verification in a secure environment. No current AI system can physically manipulate objects, manage access to physical spaces, or perform in-person security protocols. |
| Task automatability | claude-sonnet-5 | 1/5 | This task requires physical presence to hand over keys and physically admit customers into a secured vault area, which current AI systems cannot perform end-to-end.dependent on physical action. The task inherently involves a physical, in-person security interaction that AI software cannot substitute. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Banking regulations require licensed personnel to manage safe-deposit access and physical vault operations. Chain-of-custody, identity verification, and regulatory compliance mandate human accountability and sign-off on these security-critical functions. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Bank security protocols, identity verification requirements, and liability concerns around vault access create strong organizational and regulatory barriers to any automated substitution of this physical admission process. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | There is no AI alternative to the physical execution of this task, so cost comparison is not applicable. Any automation would require robotics or other hardware that remains far more expensive than employing a clerk for these administrative duties. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI-based substitute performing this physical task, so no cost comparison favors AI; a human plus physical security infrastructure is the only current cost path. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product can physically issue keys or manage physical vault access. While identity verification components exist, the core requirement of physical key issuance and vault admission cannot be performed by software systems today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product handles physical key issuance or physical vault admission; this remains entirely a human/physical-hardware function with no robotic or software product performing it in production. |
Related occupations — Office & Administrative Support
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.