Executive Secretaries and Executive Administrative Assistants
43-6011.00Median wage $76,590/yr459,910 employed (US)Rank #34 of 923 scored · top 4% by substitution
Provide high-level administrative support by conducting research, preparing statistical reports, and handling information requests, as well as performing routine administrative functions such as preparing correspondence, receiving visitors, arranging conference calls, and scheduling meetings. May also train and supervise lower-level clerical staff.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution63
Exposure61
Augmentation80
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
22 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
45%
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%62
panel mean rating 3.5/5 → substitution pressure 62/100
Technical feasibility todayw 20%59
panel mean rating 3.4/5 → substitution pressure 59/100
Cost vs. human wagew 15%68
panel mean rating 3.7/5 → substitution pressure 68/100
Adoption barriersw 20%inverted — strong barriers lower the score68
panel mean rating 2.3/5 (barrier strength) → substitution pressure 68/100
Sector adoption velocityw 10%61
panel mean rating 3.4/5 → substitution pressure 61/100
Task breakdown (22 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.
Open, sort, and distribute incoming correspondence, including faxes and email.
92CI 87–97 · exposure 95 · augmentation 88 · importance 4.0/5 · click for rater detail
Open, sort, and distribute incoming correspondence, including faxes and email.
92| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | High adoption in white-collar and professional-services sectors where email management automation and intelligent inbox tools are commonplace. Email/communication automation is among the fastest-growing clerical automation use cases. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative and professional services sectors have rapidly adopted email filtering, smart inboxes, and AI assistants for correspondence management. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments productivity by pre-sorting, flagging urgent items, suggesting routing, and reducing human review time. A human can oversee and refine the system's decisions, substantially raising effective throughput. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools (smart categorization, priority inbox, auto-drafted responses) substantially boost an assistant's efficiency in managing and distributing correspondence while a human remains in oversight for judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Current AI systems can fully automate email receipt, sorting by sender/topic/priority, and rule-based routing with >50% time savings. Fax handling via OCR and routing is also mature. This task requires minimal judgment and meets the threshold for end-to-end automation. |
| Task automatability | claude-sonnet-5 | 5/5 | Sorting and routing correspondence (email/faxes) via rules, classifiers, and AI triage is well within current off-the-shelf capability, easily meeting the 50% time-saving bar. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal legal or authorization barriers exist for automating mail sorting. Some organizations prefer human review for sensitive correspondence, but no licensing requirement mandates human involvement in this routine clerical task. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, legal, or liability barriers prevent automating correspondence sorting; it's a purely administrative function. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Email automation and sorting are commodity operations; inference costs are negligible (pennies per hundred emails) compared to loaded human wages for clerical work, making AI orders of magnitude cheaper all-in. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated email sorting and distribution costs a tiny fraction of a cent per message compared to human labor time spent manually triaging correspondence. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed products (email management automation, intelligent inbox systems, RPA tools) reliably perform classification, sorting, and distribution in production across organizations. These systems are widely available and operationally proven. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Email filtering, categorization, and routing tools (e.g., Outlook/Gmail rules, AI-based triage assistants) are widely deployed in production, though physical mail/fax handling still requires some human intervention or scanning integration. |
Manage and maintain executives' schedules.
87CI 79–95 · exposure 87 · augmentation 88 · importance 4.4/5 · click for rater detail
Manage and maintain executives' schedules.
87| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Calendar AI is already deeply embedded in enterprise information technology and professional services. Cloud calendar platforms with AI scheduling features are standard in Fortune 500 and mid-market firms, showing rapid, widespread production adoption. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Scheduling AI is widely adopted in professional services, tech, and finance sectors with mature integrations into Outlook/Google Workspace, though smaller firms lag. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | Even in a human-led model, AI scheduling tools greatly enhance an executive assistant's productivity by auto-proposing times, flagging conflicts, and handling routine requests, allowing the human to focus on judgment-heavy aspects like complex stakeholder negotiations. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI scheduling tools substantially boost assistant productivity by automating routine back-and-forth, conflict detection, and rebooking while the assistant retains oversight for judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Calendar management is highly structured, rule-based work with clear inputs (meeting requests, constraints, preferences) and outputs (scheduled time blocks). Current AI systems (calendar integration, scheduling agents, LLM-based assistants) can handle meeting requests, conflict resolution, time-zone management, and preference application with >50% time savings and quality comparable to human scheduling. |
| Task automatability | claude-sonnet-5 | 4/5 | AI scheduling assistants can autonomously handle calendar management, meeting coordination, and rescheduling with minimal human input, meeting the 50% time-saving bar for most routine scheduling work.calendar. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Although executives may prefer personal rapport and high-touch scheduling, there are no legal, licensing, or regulatory barriers to AI calendar management. Organizational friction is low; adoption is blocked mainly by habit and desire for a human touch rather than hard rules. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human handle scheduling; the main friction is executive preference for a trusted human gatekeeper and occasional sensitive judgment calls. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI scheduling costs pennies per month in SaaS or API calls, while human executive assistants cost $50k–$80k+ annually fully loaded. The cost ratio is orders of magnitude in AI's favor. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI scheduling tools cost a few dollars to tens of dollars monthly per user versus the substantial loaded cost of executive assistant time spent on calendar management, an order-of-magnitude difference. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple mature products perform this at scale in production: Google Calendar, Microsoft Outlook, and specialized executive assistant tools with AI agents already manage schedules reliably for thousands of executives with high reliability and minimal error rates. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like Clara, Reclaim.ai, Microsoft Copilot, and Google Calendar AI features are deployed at scale and reliably handle scheduling, though edge cases (VIP prioritization, complex negotiations) still require human oversight. |
Make travel arrangements for executives.
84CI 75–92 · exposure 87 · augmentation 88 · importance 4.2/5 · click for rater detail
Make travel arrangements for executives.
84| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Enterprise travel automation is widespread and maturing rapidly. Fortune 500 firms routinely deploy agentic travel systems; adoption is fast in white-collar professional services and information sectors where this role concentrates. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and professional services functions have seen fast adoption of AI scheduling and travel tools, with many firms already using automated travel management systems. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI dramatically augments an assistant's productivity by handling routine multi-step bookings, real-time price monitoring, and itinerary conflicts in seconds while the human focuses on exceptions, special requests, and relationship management with executives. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI travel tools significantly speed up itinerary building, price comparison, and rebooking while the assistant retains final judgment on executive-specific needs and exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Making travel arrangements (booking flights, hotels, ground transport, managing itineraries) is highly structured and algorithmic. Current AI agents can autonomously access booking systems, compare options, execute reservations, and handle rebooking—delivering ≥50% time savings at equal or better quality than manual booking. |
| Task automatability | claude-sonnet-5 | 4/5 | Booking travel involves well-structured, rule-based steps (comparing flights, hotels, calendars) that current AI agents and travel-management tools can largely handle with minimal human correction, though preference nuances and last-minute changes still require oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal legal or licensing barriers exist; travel booking is not a regulated profession. However, some executives prefer human coordination for complex multi-leg trips or last-minute changes, and organizational policy may require human sign-off—creating moderate adoption friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human book travel; the main friction is organizational trust, executive preference for a known human assistant, and handling exceptions/VIP treatment. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI travel automation costs pennies per booking (API calls, inference, minimal oversight) versus the loaded wage of an executive assistant (\$60k–80k annually) performing the same task. The cost ratio is 1–2% of human labor cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-driven booking tools cost a small fraction of an assistant's hourly wage for the same volume of bookings, though initial integration with corporate travel policies and executive preferences adds some overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple mature products (Concur, TravelBank, custom agentic systems) reliably automate travel booking in production at scale across enterprises. These systems demonstrably execute flight/hotel reservations, itinerary consolidation, and expense compliance without material error in daily use. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Corporate travel platforms (e.g., TripActions/Navan, Concur with AI features) already automate booking based on policy and calendar integration in production at many companies, though executive-level personalization often still involves human assistants for edge cases. |
Prepare responses to correspondence containing routine inquiries.
81CI 79–84 · exposure 75 · augmentation 100 · importance 4.0/5 · click for rater detail
Prepare responses to correspondence containing routine inquiries.
81| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Corporate and professional service sectors are rapidly adopting AI email assistants and drafting tools; major vendors (Microsoft 365 Copilot, Google Workspace features) are shipping this capability. Early adoption in larger firms is visible and accelerating. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative support functions in professional services and corporate settings have seen fast uptake of AI email and writing assistants embedded in standard productivity suites. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI-assisted response drafting dramatically increases secretary productivity on correspondence, offering templates, tone suggestions, and rapid first drafts while the human retains final judgment. This is a high-productivity-gain augmentation scenario with human in the loop. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI drafting tools substantially speed up composing routine replies while the secretary retains control over sending and personalization, a clear productivity transformation. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI language models can generate competent responses to routine correspondence (FAQs, standard requests) with 50%+ time savings, though human review and personalization are often needed. Most routine inquiry responses are templatable and within current LLM capabilities. |
| Task automatability | claude-sonnet-5 | 4/5 | Drafting responses to routine, formulaic inquiries is well within current LLM capability, requiring only final human review before sending, which meets the 50% time-saving bar for most such correspondence. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard barriers exist; routine correspondence does not require licensure or legal signoff. Organizational friction and preference for human tone/judgment offer modest friction, but no regulatory or liability barrier prevents automation of templatable inquiries. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or regulatory requirement mandates a human draft routine correspondence; organizational norms pose minimal friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | API costs for LLM-based correspondence generation are typically less than $0.01 per response, while loaded human secretary wage runs $25–40/hour for work taking 5–10 minutes; AI is 50–100x cheaper per routine response. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI drafting costs fractions of a cent per response versus minutes of skilled administrative labor, making the cost differential very large even after light human review. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature email and document automation products (including AI-assisted drafting in Outlook, Gmail, and specialized business software) reliably handle routine correspondence drafting in production. Error rates on straightforward inquiries are low enough for practical deployment. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Email drafting assistants (e.g., Gmail Smart Reply/Gemini, Outlook Copilot) are deployed at scale and reliably draft routine responses, though executives often still edit for tone and specificity. |
Compile, transcribe, and distribute minutes of meetings.
81CI 79–84 · exposure 75 · augmentation 88 · importance 3.6/5 · click for rater detail
Compile, transcribe, and distribute minutes of meetings.
81| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Tech-forward sectors (finance, consulting, software) are rapidly adopting AI meeting transcription and summarization in production; even mid-market organizations are piloting or rolling out these tools. Adoption is faster in digitalized, information-centric industries. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Meeting transcription/summarization AI is one of the fastest and most broadly adopted enterprise AI use cases across corporate and professional services sectors via Zoom, Teams, and Google Workspace integrations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI transcription and summarization dramatically augment the secretary's ability to capture complete, organized meeting records and free them for higher-judgment tasks like follow-up coordination and tone adjustment. The human remains in the loop to ensure accuracy and context. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially transforms this task by auto-generating draft minutes and action items from recordings, letting the assistant focus on review, formatting, and distribution rather than manual transcription. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can now reliably transcribe meeting audio, extract key decisions and action items, format them into structured minutes, and distribute them—often with minimal human review. Voice-to-text and LLM-based meeting summarization meet the 50% time-saving bar for well-structured meetings, though complex jargon or poor audio may still require human touch-up. |
| Task automatability | claude-sonnet-5 | 4/5 | Transcription and summarization of meeting audio/notes into structured minutes is now well within capability of AI transcription plus LLM summarization tools, meeting the 50% time-saving bar for most straightforward meetings, though distribution and nuanced editing may still need human review. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers block this automation; however, some organizations require human review for accuracy, confidentiality concerns may slow adoption, and executives may prefer a human aide for context-setting or note-taking nuance. These are soft adoption frictions rather than hard blockers. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, regulatory, or liability barriers prevent AI-generated minutes; organizations already widely adopt these tools without legal restriction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered transcription and meeting summarization costs pennies per meeting versus tens of dollars in human labor. The cost ratio is at least an order of magnitude in AI's favor, especially for routine meetings. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI transcription and summarization subscriptions cost a few dollars per user monthly versus the hourly loaded wage of an executive assistant doing manual transcription, an order-of-magnitude cost difference. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed products (Otter.ai, Fireflies.ai, Microsoft Copilot in Teams, Google Meet transcription) perform meeting transcription and summarization in production at scale with acceptable accuracy. Some products still require manual review for final accuracy and tone, but the core automation is reliably available. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products like Otter.ai, Microsoft Teams/Copilot, and Zoom AI Companion reliably transcribe and generate meeting summaries/minutes in production today, though accuracy varies with audio quality and technical jargon. |
Answer phone calls and direct calls to appropriate parties or take messages.
77CI 76–79 · exposure 75 · augmentation 63 · importance 4.1/5 · click for rater detail
Answer phone calls and direct calls to appropriate parties or take messages.
77| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | AI call handling and voicemail systems see rapid adoption in professional services, finance, and mid-to-large enterprises; many organizations have already replaced or augmented human receptionists with AI agents. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative/office support functions are adopting AI call-handling tools steadily but executive assistant roles specifically still favor human judgment for high-stakes triage, so adoption is moderate rather than fast. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists remaining humans by pre-screening calls, transcribing messages, and proposing routing, reducing cognitive load; however, the task is so amenable to full automation that augmentation is secondary to replacement. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI call transcription, screening, and message drafting significantly boost an assistant's efficiency in handling call volume while the human remains in control of judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Modern AI systems can handle call routing and message-taking with high fidelity using natural language understanding and structured routing tables. While context-dependent call classification and complex human interactions remain imperfect, current IVR systems and AI agents achieve 50%+ time savings on message capture, routing decisions, and initial triage without degrading quality. |
| Task automatability | claude-sonnet-5 | 4/5 | Modern AI voice agents can answer calls, understand intent, route to appropriate parties, and take structured messages with high accuracy, meeting the time-saving threshold for most routine call volume. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers prevent automation—call routing is not a licensed activity. Minor friction exists from organizational custom (executive preference for human contact) and integration with legacy phone systems, but nothing structurally blocks adoption. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human answer calls; main friction is executive preference for personalized service and trust with sensitive contacts. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Cloud-based AI phone systems cost $10–50/month per line and handle thousands of calls; the equivalent human labor at loaded wage is orders of magnitude higher, making AI at least 10–100× cheaper per task-equivalent. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI voice agent services cost a small fraction per call/minute compared to a human executive assistant's loaded salary for the same volume of call handling. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products like AI voicemail systems (e.g., Google Voice, Vonage, Amazon Connect) reliably perform call answering, routing, and message transcription in production at scale, though some edge cases and misroutes occur in complex organizational hierarchies. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | AI receptionist and call-routing products (e.g., AI phone agents integrated with PBX/VoIP systems) are deployed in production at many businesses today, though executive-level nuance (VIP callers, sensitive routing judgment) still requires oversight. |
File and retrieve corporate documents, records, and reports.
77CI 75–80 · exposure 75 · augmentation 75 · importance 3.8/5 · click for rater detail
File and retrieve corporate documents, records, and reports.
77| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Document management and AI-assisted retrieval are rapidly adopted in corporate sectors (finance, professional services, legal, healthcare compliance). Major enterprises have deployed or are deploying automated document processing pipelines, making this a fast-moving, information-heavy domain. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Corporate and professional services environments have widely adopted digital document management and AI search tools, making this a well-established automation pattern in white-collar administrative work. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists executives and administrative staff through intelligent search, auto-tagging, and retrieval suggestions that dramatically reduce time to locate documents. The human retains judgment over sensitivity, retention policies, and complex filing decisions while AI handles the bulk of the mechanical work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI search, auto-tagging, and natural-language query tools significantly speed up how executive assistants locate and organize documents, even where full replacement hasn't occurred for edge cases. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can handle a substantial portion of this task—document classification, OCR, metadata extraction, and retrieval via semantic search or database queries are mature. End-to-end automation of filing and retrieval (excluding edge cases) meets the ≥50% time-saving threshold with off-the-shelf tools, though some manual oversight and exception handling remains. |
| Task automatability | claude-sonnet-5 | 4/5 | Modern document management systems and AI-powered search/retrieval tools can automate most of the filing and retrieval workflow for digital documents, though initial setup and handling of physical or legacy records limit full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard regulatory or licensing barriers exist for automated filing and retrieval of corporate documents; however, some organizations impose data governance policies, audit trails, and human sign-off for sensitive records, creating modest organizational friction but not legal prohibition. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or regulatory requirement mandates a human perform basic filing and retrieval; it's a purely administrative function with minimal friction to automate. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven document management and retrieval (cloud storage, automated indexing, search APIs) costs a fraction of full-time administrative labor for equivalent throughput. A secretary's salary easily exceeds the monthly cost of enterprise document systems handling thousands of filings and retrievals. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Cloud storage and AI-assisted indexing/retrieval cost a fraction of a cent per document versus staff time spent manually filing and searching, though integration and migration costs offset some savings initially. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (document management systems, RPA, AI-assisted search) reliably perform parts of filing and retrieval in production at scale across organizations. Systems like Salesforce, SharePoint with AI indexing, and dedicated document-capture platforms demonstrate mature capability, though some customization and human validation are typically required. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Enterprise document management platforms (SharePoint, Google Workspace, DMS with AI search like M-Files) reliably index, tag, and retrieve corporate documents in production today, though physical records and inconsistent legacy filing still require human handling. |
Prepare invoices, reports, memos, letters, financial statements, and other documents, using word processing, spreadsheet, database, or presentation software.
77CI 75–79 · exposure 75 · augmentation 100 · importance 4.2/5 · click for rater detail
Prepare invoices, reports, memos, letters, financial statements, and other documents, using word processing, spreadsheet, database, or presentation software.
77| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Document automation is rapidly adopted across professional services, finance, and corporate sectors, with widespread deployment of AI-assisted office tools and RPA in administrative workflows. Adoption is already substantial and accelerating. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative functions in professional services and corporate settings are adopting generative AI tools for document drafting quickly, driven by major software vendors embedding AI directly into productivity suites. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI assistants significantly raise human productivity by drafting documents, formatting, and populating fields, while humans retain control over accuracy, tone, and final sign-off, creating a clear augmentation dynamic. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI drastically speeds up drafting of routine documents while executive assistants retain control over final content, distribution, and judgment calls, making this a strong augmentation use case. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably generate invoices, reports, memos, letters, and financial statements from templates and structured data using word processing and spreadsheet tools, achieving substantial time savings. Template-based document generation is mature and widely deployed, though some customization and quality checks may still require human oversight. |
| Task automatability | claude-sonnet-5 | 4/5 | Drafting invoices, memos, letters, and standard reports from templates or data inputs is well within current LLM and office-suite AI capabilities, saving significant drafting time, though final review/formatting nuances still need human input. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers prevent automation of routine document preparation; organizations mainly face internal governance friction around document sign-off and liability for errors in financial statements, which is organizational rather than regulatory. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for drafting these documents, though financial statements may need accountant sign-off and some organizations prefer human-reviewed correspondence for tone/confidentiality. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference and integration costs for document generation are negligible compared to the loaded hourly wage of an executive secretary, making automated generation orders of magnitude cheaper per document produced. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted document generation costs a small fraction of a cent to a few cents per document versus the loaded cost of executive assistant time, though integration and review add some overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple production systems (Microsoft Office automation with AI copilots, Google Workspace AI, specialized business intelligence tools) demonstrably perform document generation at scale in organizations. Error rates on routine documents are low, though complex financial statements with novel requirements may need human review. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Copilot, Google Workspace AI, and similar tools are deployed at scale in real organizations to draft documents, fill templates, and generate reports from data, though accuracy on complex financial statements still requires oversight. |
Process payroll information.
76CI 70–81 · exposure 75 · augmentation 50 · importance 4.3/5 · click for rater detail
Process payroll information.
76| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Payroll automation has achieved deep, rapid adoption across information, finance, and professional-services sectors; most mid-to-large organizations have already shifted to automated payroll systems, making this one of the earliest and most mature automation domains in the workplace. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Payroll software adoption is already widespread and mature across nearly all firm sizes and sectors, representing one of the most deeply automated back-office functions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | While AI handles core payroll tasks, human executive assistants still use AI-powered dashboards, alerts, and reporting tools to monitor exceptions, reconcile discrepancies, and make decisions on unusual compensation cases, providing meaningful productivity improvement without full replacement. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Where full automation isn't in place, AI-enabled payroll tools assist with error-checking, anomaly detection, and streamlining data entry, improving efficiency for the human handling exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Payroll processing is highly structured, rule-based work involving data entry, calculation, tax computations, and report generation—all tasks where AI and RPA systems can achieve significant time savings. Current systems can extract data, compute withholdings, generate checks, and file reports with minimal human oversight, easily exceeding the 50% threshold. |
| Task automatability | claude-sonnet-5 | 4/5 | Payroll data entry, calculations, and processing are largely rule-based and already handled by payroll software with minimal human input beyond exception handling and approvals. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Payroll has moderate barriers: regulatory compliance requirements (tax filing, wage-and-hour laws) and audit trails create oversight obligations, and many organizations prefer human sign-off on payroll for liability and control reasons. However, no licensing requirement explicitly forbids AI payroll automation, and the barriers are primarily organizational friction rather than legal prohibition. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Some compliance and audit requirements exist around payroll accuracy and tax law, but no licensing requirement mandates a human execute the mechanical processing itself. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated payroll processing costs pennies to dollars per employee per pay cycle, versus an executive assistant's fully loaded hourly rate applied to payroll tasks; the cost ratio is easily an order of magnitude in favor of automation. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated payroll processing costs a small fraction per employee compared to manual processing time by an executive assistant, though software licensing and setup add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature payroll software (ADP, Workday, BambooHR) with embedded AI/automation features are deployed at scale in organizations worldwide and reliably process payroll end-to-end. While some edge cases (unusual deductions, multi-state/multi-country rules) require human review, the core process is production-proven. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mature payroll platforms (ADP, Gusto, Workday) reliably automate calculations, tax withholding, and disbursement at scale in production today, though edge cases still need human review. |
Prepare agendas and make arrangements, such as coordinating catering for luncheons, for committee, board, and other meetings.
73CI 67–79 · exposure 70 · augmentation 100 · importance 3.9/5 · click for rater detail
Prepare agendas and make arrangements, such as coordinating catering for luncheons, for committee, board, and other meetings.
73| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Large organizations and professional services firms are actively adopting calendar and meeting-coordination agents; this is a core use case for enterprise AI assistants and executive productivity tools already in production. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative/professional services functions are moderate-to-fast adopters of AI scheduling and drafting tools, though full replacement of logistics coordination lags behind pure text tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI excels at real-time agenda updates, catering matching, calendar conflict resolution, and draft agenda generation; a human executive remains in the loop for final decisions, resulting in significant productivity gains while maintaining human oversight. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially speeds up agenda drafting, meeting prep, and logistics coordination (emails, reminders, catering orders) while the human assistant remains responsible for judgment calls and relationship management. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Calendar integration, catering vendor lookup, dietary tracking, and agenda assembly are straightforward informational tasks that current AI can execute end-to-end with significant time savings. Task requires some human judgment on formatting and tone, but the core work—scheduling coordination, vendor contact, and document assembly—is substantially automatable. |
| Task automatability | claude-sonnet-5 | 4/5 | AI scheduling assistants and LLMs can draft agendas from meeting context and coordinate catering orders via integrated tools, though final confirmation and relationship management with vendors still often needs human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No licensing requirement; users expect human secretaries for meeting prep but organizational friction remains low. Oversight and executive preference for human touch provide mild friction, but nothing legally restricts automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists, but organizational trust, confidentiality of board materials, and preference for a known human point of contact for external caterers create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of API calls and LLM inference for calendar access, vendor research, and email composition is negligible compared to the loaded labor cost (salary + benefits) of a human assistant performing these coordination tasks. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Drafting agendas and routine logistics coordination via AI tools costs a small fraction of an executive assistant's loaded wage, though some human verification and vendor communication still adds cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Current AI tools (calendar APIs, email automation, document generation via agents) are deployed in production for similar coordination tasks. Existing calendar assistants and workflow platforms handle meeting scheduling reliably; catering vendor matching and arrangement confirmation remain partially manual but achievable with agent-based workflows at scale. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI scheduling tools (e.g., calendar assistants, agenda-generation via LLMs) exist and are deployed, but full end-to-end catering coordination and nuanced agenda-setting for sensitive board matters is not yet a mature, widely-deployed single product. |
Perform general office duties, such as ordering supplies, maintaining records management database systems, and performing basic bookkeeping work.
67CI 55–79 · exposure 62 · augmentation 75 · importance 3.9/5 · click for rater detail
Perform general office duties, such as ordering supplies, maintaining records management database systems, and performing basic bookkeeping work.
67| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Large enterprises and mid-market organizations have rapidly adopted automation tools for administrative tasks, with RPA and AI-powered administrative platforms now common in corporate and professional services sectors. Adoption is measurable and accelerating. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Office administration functions across many sectors are adopting cloud-based bookkeeping and procurement automation at a moderate pace, with pilots and partial deployments common but full replacement still rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems significantly augment human administrative assistants by automating routine ordering, flagging database anomalies, and pre-filling bookkeeping entries, allowing humans to focus on strategic support, stakeholder management, and exception-handling. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered tools significantly speed up bookkeeping entries, database maintenance, and supply reordering, letting the assistant focus on higher-judgment tasks while still reviewing outputs. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Most components—ordering supplies via automated systems, database record maintenance, and routine bookkeeping—are highly automatable with modern AI and RPA tools achieving well over 50% time savings. The task requires minimal judgment and involves mostly structured, repetitive processes that AI can execute end-to-end. |
| Task automatability | claude-sonnet-5 | 3/5 | Portions like data entry, supply ordering via templates, and basic bookkeeping can be automated with existing software integrations, but overall task requires coordination and judgment that still needs human oversight for full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist; automation is permitted without licensing requirements. Minor friction includes need for human oversight of financial records and some organizational preference for human judgment on non-standard requests, but these do not prevent substantial automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human for these administrative tasks, though organizational habits and trust in accuracy for bookkeeping create some friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven tools for supply ordering, database management, and bookkeeping cost orders of magnitude less than human labor once amortized over transaction volume. A single system can handle the work of multiple administrative staff simultaneously. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Software subscriptions and automation tools are cheaper than a dedicated staff hour for repetitive tasks, but implementation, maintenance, and oversight costs keep the ratio only moderately favorable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (automation software, inventory management systems, accounting software with AI modules) reliably handle supply ordering, records management, and bookkeeping in production environments today. Some edge cases and policy variations require human review, but core execution is demonstrably feasible at scale. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like QuickBooks automation, procurement platforms, and database tools exist and are used in production, but integrating them into a seamless workflow for a specific office still requires human setup and correction. |
Provide clerical support to other departments.
66CI 57–75 · exposure 62 · augmentation 88 · importance 3.6/5 · click for rater detail
Provide clerical support to other departments.
66| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Information-sector organizations, professional services, and corporate offices have rapidly adopted AI-driven scheduling, document management, and workflow automation over the past 2–3 years; pilots and production deployments are increasingly common in digitized enterprises. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and office support functions in professional services and corporate settings are among the faster-adopting areas for AI copilots and workflow automation tools. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants substantially augment executive secretaries by automating routine scheduling, prioritizing emails, generating meeting notes, and organizing documents, allowing them to focus on relationship management, strategic coordination, and complex stakeholder communication. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially boosts productivity for clerical support tasks like drafting, scheduling, and information routing while the human assistant retains oversight and judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Routine clerical support tasks like scheduling, email management, document organization, and filing can be substantially automated with current AI and workflow tools, but complex coordination, stakeholder management, and judgment calls still require human oversight. Approximately 50% of routine clerical work (data entry, calendar management, basic correspondence) meets the automation threshold. |
| Task automatability | claude-sonnet-5 | 4/5 | Generic clerical tasks (scheduling, correspondence, data entry, document formatting) are well within current AI capabilities using off-the-shelf tools, though some in-person coordination and judgment calls remain. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard legal or licensing barriers exist for automating clerical work; barriers are primarily organizational (change resistance, need for human judgment on sensitive matters, preference for personal assistant continuity) rather than regulatory or liability-driven. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement and minimal regulatory barrier exists for clerical work, though some organizational trust and confidentiality concerns create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI automation of routine clerical tasks (scheduling, filing, correspondence templates) costs roughly comparable to a junior administrative assistant wage after accounting for integration, maintenance, and error correction, but does not yet reach an order-of-magnitude cost advantage for the full task scope. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-based clerical tools (calendar bots, transcription, templated correspondence) cost a small fraction of a human assistant's loaded wage for equivalent throughput on routine subtasks. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Multiple products (calendar automation, document management systems, email filtering, RPA tools) perform clerical support functions in production, but they typically require significant setup, human oversight for edge cases, and integration with existing systems. Narrow scope and error rates on context-dependent tasks limit maturity. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | AI scheduling assistants, email drafting tools, and document automation are already deployed widely in production for clerical support functions across many organizations. |
Set up and oversee administrative policies and procedures for offices or organizations.
62CI 38–87 · exposure 58 · augmentation 88 · importance 3.7/5 · click for rater detail
Set up and oversee administrative policies and procedures for offices or organizations.
62| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Mid-to-large organizations in professional services, finance, and information sectors are actively adopting workflow automation and AI-assisted policy generation; smaller organizations and government lag, resulting in solid but not universal deployment. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and professional services sectors show moderate AI adoption for document drafting and workflow tools, though full policy oversight automation is not yet in production use. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically assists by generating first drafts, flagging compliance gaps, and maintaining policy consistency across departments; an executive secretary using these tools can oversee and customize procedures for multiple offices in the time previously spent on one, multiplying their effectiveness. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by drafting policy language, summarizing best practices, and flagging compliance issues, significantly speeding up the administrative assistant's work while they retain oversight. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | AI can now draft, review, and implement administrative policies and procedures at scale using document generation, template systems, and workflow automation; a human executive can oversee final approval in minutes rather than hours of manual drafting and coordination, easily meeting the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | This task requires organizational judgment, stakeholder negotiation, and contextual understanding of office culture that current AI cannot fully replicate end-to-end, though drafting policy documents can be assisted.the design and oversight components resist full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Policy and procedure setup typically require organizational approval and sign-off by leadership rather than external licensing; modest friction exists around organizational change management and stakeholder buy-in, but no legal mandate requires a human to author or own the task itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There are no licensing requirements for this role, but organizational trust, authority, and accountability for setting binding office policy create moderate friction against full AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered policy drafting and procedure templating cost pennies per policy versus weeks of high-wage executive assistant time ($25–40/hour fully loaded); the cost differential is at least an order of magnitude in favor of AI. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Because the core oversight and decision-making work still requires a human administrator, AI only reduces drafting time, so the all-in cost including human oversight remains comparable to a human doing the full task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products exist (Microsoft 365 automation, process mining software, policy-generation templates with AI assistance) that reliably draft and organize administrative procedures in production; minor gaps remain in handling highly customized organizational cultures or edge-case regulatory nuances. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | No deployed product autonomously sets up and oversees administrative policy systems; AI tools exist for drafting text but not for the ongoing oversight and enforcement function this task implies. |
Supervise and train other clerical staff and arrange for employee training by scheduling training or organizing training material.
60CI 32–87 · exposure 58 · augmentation 75 · importance 3.4/5 · click for rater detail
Supervise and train other clerical staff and arrange for employee training by scheduling training or organizing training material.
60| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Large professional services, finance, and corporate sectors have already deployed calendar and workflow automation at scale; mid-market firms are rapidly adopting AI scheduling assistants. This is one of the earliest automated EA functions and adoption continues to accelerate. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support functions are adopting AI tools for scheduling and content creation at a moderate pace, though supervisory tasks lag behind. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants significantly augment scheduling by auto-suggesting time slots, detecting conflicts, and proactively organizing materials, allowing the human secretary to focus on judgment calls and relationship management around training design and stakeholder coordination. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can significantly help draft training materials, schedule sessions, and track completion, meaningfully boosting the administrative assistant's efficiency while they retain the supervisory role. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Scheduling training sessions can be fully automated via calendar and email systems; organizing and distributing training materials is routine document management. Both activities represent clear, repetitive administrative workflows that modern AI agents with access to email, calendars, and document systems can execute end-to-end with significant time savings. |
| Task automatability | claude-sonnet-5 | 2/5 | Scheduling and organizing training materials can be partly automated, but supervising and training staff requires interpersonal judgment, mentoring, and performance evaluation that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal regulatory or legal barriers exist for automating scheduling and training material distribution. Organizational friction around perceived loss of human coordination touch is the main barrier, though adoption is already widespread in large enterprises. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement, but organizational trust, accountability for personnel management, and the human-contact nature of supervision create moderate friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI scheduling and document management cost negligibly (fractional cents per scheduling action or email), while an executive secretary's loaded wage for this task is $25–40+ per hour. The cost ratio heavily favors automation. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply handle logistics like scheduling, but the supervisory and interpersonal aspects still require a human, so overall cost savings are limited relative to full task substitution. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Calendar scheduling and email automation are mature capabilities deployed widely in enterprise systems; document organization and distribution via AI-assisted workflows are increasingly common. Some edge cases (complex multi-stakeholder coordination, last-minute changes) remain but core scheduling and material distribution work reliably at scale. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products exist for scheduling and content generation (e.g., calendar tools, LMS platforms), but no deployed product performs the supervisory/managerial component of staff training reliably. |
Attend meetings to record minutes.
58CI 37–79 · exposure 55 · augmentation 88 · importance 3.5/5 · click for rater detail
Attend meetings to record minutes.
58| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Many organizations have piloted meeting transcription and AI summarization, but most still rely on human secretaries to attend meetings and produce final minutes; adoption is growing but unevenly distributed. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Meeting AI assistants have seen rapid uptake in corporate and professional service settings, integrated directly into widely used platforms like Zoom, Teams, and Google Meet. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI transcription and summarization substantially assist secretaries by reducing manual transcription time and providing initial drafts, allowing them to focus on accuracy, context, and distributing polished minutes quickly. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI note-taking tools substantially reduce the burden of manual minute-taking, letting assistants focus on verification, formatting, and follow-up actions rather than real-time transcription. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can transcribe audio and summarize meetings, but capturing accurate context, decisions, and action items typically requires human judgment and real-time understanding of organizational priorities. Current systems often miss nuance and require substantial post-processing. |
| Task automatability | claude-sonnet-5 | 4/5 | AI transcription and summarization tools can capture audio, transcribe speech, and generate structured minutes with high accuracy, meeting the 50% time-saving bar for most standard meetings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Organizational norms, executive preference for human presence, and the need for real-time note-taking judgment create friction, though no formal licensing barrier prevents automation of the recording itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for minute-taking, though some organizations have confidentiality concerns, need for human judgment on what's material, and preference for a trusted human presence in sensitive meetings. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI transcription and summarization tools reduce time investment, but the loaded cost of an executive secretary plus the cost of AI systems and oversight remains lower than replacing the role entirely; savings are partial, not transformative. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Subscription-based transcription/summarization tools cost a few dollars per user per month versus the substantial loaded cost of an executive assistant's time spent in meetings and drafting minutes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Transcription products (Otter, Microsoft Teams transcription) are deployed, but producing publication-ready minutes with accurate context and proper action-item attribution still requires manual review and editing by a human in most organizations. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like Otter.ai, Zoom AI Companion, and Microsoft Copilot are deployed at scale in real organizations to transcribe and summarize meetings, though accuracy varies with audio quality, accents, and cross-talk. |
Conduct research, compile data, and prepare papers for consideration and presentation by executives, committees, and boards of directors.
54CI 46–61 · exposure 50 · augmentation 88 · importance 3.9/5 · click for rater detail
Conduct research, compile data, and prepare papers for consideration and presentation by executives, committees, and boards of directors.
54| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Executive support functions in larger, digitized organizations are piloting AI-assisted research and drafting, but full end-to-end automation of board-level preparation remains uncommon. Adoption is growing in document assembly and research aggregation but remains uneven across sectors. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Executive support functions sit within white-collar/professional services and are seeing fast adoption of AI drafting and research tools in many organizations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists this task by rapidly gathering, summarizing, and organizing source material; creating draft outlines and talking points; and formatting documents—all while the executive secretary maintains editorial control and strategic alignment. This augmentation materially raises human productivity in the research and compilation phases. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools substantially speed up research gathering, data compilation, and document drafting while the executive assistant retains control over final judgment and presentation. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can automate significant portions of research compilation, data aggregation, and document formatting, but requires human oversight for framing arguments, determining relevance, and ensuring strategic alignment with executive priorities. The task involves judgment calls about what matters to a specific board that resist full automation. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can draft research summaries, compile data, and generate presentation-ready documents quickly, but curating source relevance, verifying accuracy, and tailoring to executive judgment still require human oversight, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Executives and boards typically expect a trusted human intermediary who understands context and stakes; there is implicit organizational friction and preference for human judgment on high-stakes presentations. No strict legal barrier exists, but reputational risk and internal governance norms create meaningful adoption friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for this task, but organizational trust, confidentiality concerns, and the need for judgment in framing information for executives create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | While AI inference is cheap, the task demands careful integration, fact-checking oversight, and iteration to meet board-level quality standards. The loaded cost of human review and validation offsets much of the AI efficiency gain, keeping total-cost-of-ownership only moderately better than human work. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted research and drafting is dramatically cheaper per unit of output than a human assistant's time, though some human review and integration cost remains. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Deployed products (search tools, LLMs, document generators) can assist with research and compilation, but real-world systems still require substantial human review for accuracy, tone, and fit-to-context. Material error rates remain in citation accuracy and subtle misinterpretation of nuance, limiting reliability at scale. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like Copilot, ChatGPT with browsing, and enterprise research assistants are deployed and used for drafting and compiling data, but they still produce errors, hallucinations, or require significant editing before executive-level presentation. |
Read and analyze incoming memos, submissions, and reports to determine their significance and plan their distribution.
52CI 46–59 · exposure 50 · augmentation 88 · importance 3.7/5 · click for rater detail
Read and analyze incoming memos, submissions, and reports to determine their significance and plan their distribution.
52| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Pilot adoption of AI document routing and triage is common in large, digital-heavy organizations (finance, professional services), but production deployment remains patchy and often confined to lower-stakes, high-volume incoming mail. Full-task automation adoption has not yet reached early majority status. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support functions are adopting AI copilots at a moderate pace, with pilots widespread but full автономный routing still uncommon. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists by auto-summarizing documents, flagging urgent content, suggesting recipients, and creating draft routing plans—capabilities that measurably raise human productivity without removing the executive secretary from the loop. The human remains the arbiter of significance and distribution authority. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI excels at summarizing, flagging priority items, and drafting distribution suggestions, significantly speeding up the human's review and decision process. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can extract key information, categorize documents, and suggest routing with reasonable accuracy, but determining 'significance' requires contextual judgment about organizational priorities, stakeholder relationships, and strategic implications that often demands human oversight. The task is partially automatable with setup, but full end-to-end replacement with 50% time savings at equal quality is uncertain. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can read, summarize, and triage documents effectively, but determining organizational significance and appropriate distribution requires contextual judgment about people, politics, and priorities that current systems handle imperfectly without heavy customization. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Organizational friction and preference for human judgment on sensitive/strategic documents create moderate barriers; however, there are no legal licensing requirements or hard liability rules preventing automation of routing. Most barriers are cultural—trust in AI to understand priority and discretion—rather than regulatory. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but sensitive/confidential content and trust dependencies on discretion create moderate organizational friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The technology (document processing, classification APIs) is inexpensive per unit, but integration with existing workflows, the need for human oversight of edge cases, and retraining on organizational-specific significance criteria add overhead. On balance, AI is not yet cheaper than a junior administrative assistant for this mixed task. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Once integrated, AI summarization and classification costs are far lower per document than executive assistant time, though initial setup and oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Document classification and triage products exist in production (email filtering, document management systems with routing rules), but they struggle with nuanced significance judgments and often require human review before distribution. Error rates in misclassification or missed priority items remain material enough that most organizations still rely on human screening. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Email/document triage and summarization tools (e.g., Outlook Copilot, Gmail AI features) exist in production, but reliable autonomous routing based on nuanced significance judgments is narrow and often requires human review. |
Review operating practices and procedures to determine whether improvements can be made in areas such as workflow, reporting procedures, or expenditures.
45CI 35–55 · exposure 38 · augmentation 75 · importance 3.3/5 · click for rater detail
Review operating practices and procedures to determine whether improvements can be made in areas such as workflow, reporting procedures, or expenditures.
45| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Administrative functions and process improvement are evolving toward AI-assisted workflows, but most organizations still rely on manual review by experienced staff; deployment remains sporadic rather than mainstream. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office professional roles are seeing moderate AI tool adoption (reporting dashboards, analytics copilots) but full process-review automation remains uncommon in practice. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially augment this task by rapidly analyzing large volumes of procedural documentation, identifying patterns, and generating preliminary recommendations for inefficiency, allowing the executive secretary to focus on validation, stakeholder input, and change management. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist by analyzing reports, summarizing inefficiencies, and drafting recommendations, significantly speeding up the assistant's review process while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can assist with document review, process mapping, and identifying efficiency gaps from existing procedures and data, but the task requires judgment about organizational context, stakeholder concerns, and implementation feasibility that typically demands human oversight and decision-making. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can help analyze workflow data and suggest improvements, but the full task requires organizational judgment, stakeholder interviews, and contextual understanding that current systems cannot autonomously perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No legal requirement mandates that a human must conduct this task, though organizational hierarchy and decision-making authority over procedural changes create friction; executives and management typically retain final judgment on implementation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing barrier exists, but organizational trust, need for contextual judgment, and internal politics create moderate friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 3/5 | AI-assisted analysis of procedures could reduce manual document review time, but the value proposition depends on integration with existing systems and the need for experienced human judgment; costs are roughly comparable to a portion of a junior analyst's time. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Effective review requires significant human oversight, data gathering, and interpretation; AI tools reduce some analysis time but don't yet replace the human cost of judgment-heavy review at scale. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Products exist (workflow analysis tools, document AI, process mining software) that can extract and summarize procedural information and flag inefficiencies, but reliable end-to-end improvement recommendations require human validation and understanding of organizational nuance. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some analytics and process-mining tools exist to flag inefficiencies, but no deployed product independently conducts full operating-practice reviews and recommends actionable changes reliably in production. |
Coordinate and direct office services, such as records, departmental finances, budget preparation, personnel issues, and housekeeping, to aid executives.
38CI 30–46 · exposure 38 · augmentation 75 · importance 4.1/5 · click for rater detail
Coordinate and direct office services, such as records, departmental finances, budget preparation, personnel issues, and housekeeping, to aid executives.
38| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption remains uneven; while large corporate finance teams have automated accounting processes, smaller and mid-market firms lag significantly. The human relational and judgment aspects of executive support slow full replacement; AI adoption here is mostly piecemeal tool integration rather than cohesive agent displacement. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support functions in professional services are adopting AI tools for scheduling, drafting, and reporting at a moderate pace, though full coordination roles remain human-led. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | Current AI systems (document automation, calendar management, draft budget summaries, personnel data dashboards) substantially aid human assistants in efficiency and accuracy. AI-assisted drafting and real-time alerts on compliance or scheduling conflicts meaningfully lift productivity while the assistant remains the decision-maker and coordinator. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools substantially help with budget analysis, document drafting, scheduling, and data organization, meaningfully boosting the productivity of the person performing this coordination role. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Portions of this task (records management, budget preparation, some personnel tracking) are automatable with current systems, but coordination and executive judgment remain human-dependent. Reaching the 50% time-saving threshold requires integrating multiple AI tools (document management, financial processing, scheduling) with significant setup and workflow redesign. |
| Task automatability | claude-sonnet-5 | 2/5 | This is a broad coordination and oversight bundle spanning multiple domains (records, finance, personnel, facilities) requiring judgment, prioritization, and interpersonal coordination that current AI cannot fully replace end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Moderate barriers: executives often prefer trusted human gatekeepers for sensitive personnel and budget decisions; some organizations have compliance and audit requirements for financial sign-off that necessitate human judgment or signature. No hard legal barriers, but organizational culture and liability concerns create friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but sensitive personnel and financial decisions carry liability and confidentiality concerns, and executives typically want a trusted human in this coordinating role. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Automating scattered parts (e.g., document filing, expense categorization) costs less per task, but comprehensive office coordination automation requires multiple tool licenses, integration overhead, and ongoing human oversight. Total cost approaches or exceeds a mid-level secretary wage for the full workflow. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply handle discrete sub-tasks like data entry or scheduling, but the coordination and judgment-heavy oversight role still requires a human, so overall cost savings versus a full-time assistant are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Products exist for subcomponents (RPA for records, accounting software, HR platforms) but no single integrated system reliably handles the full coordination and judgment required. Error rates on nuanced personnel and budgeting decisions remain material; production deployment is partial and sector-dependent. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Point tools exist for sub-tasks like budget tracking or document management, but no deployed product coordinates and directs the entire multi-domain office service function autonomously. |
Interpret administrative and operating policies and procedures for employees.
36CI 30–41 · exposure 25 · augmentation 75 · importance 3.8/5 · click for rater detail
Interpret administrative and operating policies and procedures for employees.
36| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption in this administrative domain remains cautious and pilot-heavy; while large enterprises experiment with AI-driven HR support, actual production displacement of this specific task (policy interpretation) lags behind adoption in finance or sales. Most organizations still rely on human administrative staff or HR for authoritative policy guidance. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support roles are seeing moderate AI tool adoption (e.g., HR chatbots, internal knowledge assistants), but full interpretive tasks remain in pilot stages. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists by drafting clear, comprehensive policy interpretations that an executive secretary can review and refine, generating scenario-based guidance, and flagging ambiguities. The human remains in control while AI speeds up research and drafting, materially raising an assistant's productivity on this task. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can quickly retrieve relevant policy text, draft explanations, and summarize procedures, significantly aiding the secretary who then contextualizes and delivers the interpretation. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Interpreting policy and procedures involves contextual understanding and nuanced judgment about how abstract rules apply to specific scenarios, which remains difficult for current AI. While AI can retrieve and summarize policy documents, applying them to novel employee situations typically requires human discretion and organizational knowledge that AI systems lack. |
| Task automatability | claude-sonnet-5 | 2/5 | Interpreting policy requires contextual judgment, organizational knowledge, and interpersonal delivery that current AI can partially support but not fully replace end-to-end at equal quality.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Organizations face moderate friction: HR liability concerns about incorrect policy interpretation, requirement for human sign-off on sensitive guidance, and employee preference for human interaction on policy questions. However, no hard legal requirement mandates a licensed human perform this task, and organizations can substitute AI with oversight. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists, but organizational trust, nuanced judgment calls, and accountability for correct interpretation create moderate friction against full substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI solutions for policy interpretation require significant setup (knowledge base creation, model fine-tuning, oversight workflows) and ongoing human review of outputs, making them comparable to or slightly cheaper than a dedicated human, but not an order of magnitude cheaper given quality requirements. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-based Q&A tools are cheap to run, but the need for human oversight, customization, and error correction makes overall cost roughly comparable to human labor for this task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably interprets organizational policies for diverse employee queries at scale; products exist for policy document search and summarization but not for genuine interpretation and application. Error rates on boundary cases remain material, and most implementations are narrow (specific FAQ responses) rather than general policy interpretation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Chatbots and knowledge-base tools can answer some policy questions, but no deployed product reliably interprets nuanced administrative policy for employees across varied organizational contexts. |
Greet visitors and determine whether they should be given access to specific individuals.
26CI 23–30 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail
Greet visitors and determine whether they should be given access to specific individuals.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Adoption of autonomous AI visitor screening remains extremely rare in real organizations; most deployments remain pilots or research-stage, with offices retaining human receptionists for both security and service reasons. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Physical front-desk and visitor management functions are in a low-digitization, low-AI-adoption segment relative to purely digital administrative tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist a human receptionist by automatically checking calendars, flagging security alerts, or logging visitor information, raising their efficiency; however, the human remains essential for judgment and social interaction. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-driven visitor management systems, calendars, and identity verification tools can meaningfully assist an assistant in vetting and directing visitors even though final judgment remains human. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Greeting visitors and controlling access requires nuanced social judgment, real-time decision-making about who should meet whom, and handling exceptions; while AI systems could handle routine scheduling checks, the interpersonal gatekeeping role with discretionary judgment remains largely beyond current automation at equal quality. |
| Task automatability | claude-sonnet-5 | 2/5 | Physical greeting and visual/contextual judgment about access require presence and situational judgment that current AI cannot fully replicate end-to-end, though scheduling checks could be partially automated.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Legal liability (admitting unauthorized persons), security responsibility, company liability for access failures, and customer expectations for human hospitality and judgment create significant barriers; many organizations require a human to sign off on or perform access decisions. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing requirement, but liability, security, and interpersonal trust considerations create meaningful organizational friction against removing human judgment from access control. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Building and maintaining an AI reception system (hardware, software, integration, oversight) would likely exceed the loaded wage of a receptionist in most settings, especially given the need for human fallback and exception handling. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Hardware, kiosks, or badge systems plus human oversight for exceptions add cost comparable to or exceeding a shared receptionist/assistant's marginal cost for this specific duty. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed products reliably perform visitor screening and access control autonomously; systems exist for appointment scheduling or lobby check-in but lack the judgment needed to independently determine access eligibility and handle unscheduled or edge-case visitors in production settings. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some receptionist kiosks and AI check-in systems exist but are narrow, handle only routine cases, and still require human backup for judgment calls and security decisions. |
Meet with individuals, special interest groups, and others on behalf of executives, committees, and boards of directors.
8CI 0–16 · exposure 8 · augmentation 50 · importance 3.6/5 · click for rater detail
Meet with individuals, special interest groups, and others on behalf of executives, committees, and boards of directors.
8| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task involves high-stakes interpersonal and fiduciary representation, so adoption of autonomous AI substitution is essentially zero across all sectors. AI augmentation of scheduling and preparation is emerging, but replacement is not happening. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | While administrative professions see AI tool adoption for scheduling and notes, the core representative meeting function itself shows minimal displacement trends. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by preparing briefing notes, managing calendar logistics, tracking follow-ups, and summarizing outcomes, meaningfully raising an executive secretary's efficiency in coordinating and preparing for meetings. However, the core meeting function itself remains human-dependent. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help prepare briefing materials, talking points, and follow-up summaries, meaningfully aiding the secretary's effectiveness in these meetings without replacing their presence. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Only narrow parts of this task can be automated—scheduling meetings and initial information gathering are feasible, but the interpersonal negotiation, relationship-building, and real-time judgment required in actual meetings with stakeholders demand human presence and discretion. Meeting as a proxy for an executive requires nuanced understanding of organizational context and authority that current AI cannot reliably replicate. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physically or virtually representing an executive in real-time interpersonal negotiation and relationship management, which current AI cannot perform end-to-end., as it demands human presence, judgment, and trust-building. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Strong barriers exist: executives and boards require human intermediaries with personal accountability; external parties expect human representatives with implicit authority and trust; legal and fiduciary responsibility rests on the human agent, not an AI system. Organizational norms and liability structures make substitution nearly impossible. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Trust, authority delegation, confidentiality, and organizational norms create strong barriers since stakeholders expect a human representative with real decision-making credibility. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Even if the interpersonal meeting could theoretically be automated, the liability, oversight, and verification costs—plus the need for human fallback—would exceed the cost of having an executive secretary do the task, given their broad value to the organization. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this representative role, so cost comparison favors the human by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can reliably conduct representative meetings on behalf of executives with external parties; this requires legal authority, trust, and accountability that only humans can provide. AI scheduling and note-taking tools exist, but autonomous deputized meeting attendance is not a mature product capability. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product substitutes a human representative in live meetings with stakeholders on behalf of executives; this remains firmly human-performed. |
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.