Secretaries and Administrative Assistants, Except Legal, Medical, and Executive
43-6014.00Median wage $47,540/yr1,706,790 employed (US)Rank #14 of 923 scored · top 2% by substitution
Perform routine administrative functions such as drafting correspondence, scheduling appointments, organizing and maintaining paper and electronic files, or providing information to callers.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution70
Exposure67
Augmentation78
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
31 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
65%
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%67
panel mean rating 3.7/5 → substitution pressure 67/100
Technical feasibility todayw 20%68
panel mean rating 3.7/5 → substitution pressure 68/100
Cost vs. human wagew 15%74
panel mean rating 4.0/5 → substitution pressure 74/100
Adoption barriersw 20%inverted — strong barriers lower the score77
panel mean rating 1.9/5 (barrier strength) → substitution pressure 77/100
Sector adoption velocityw 10%66
panel mean rating 3.6/5 → substitution pressure 66/100
Task breakdown (31 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.
Conduct searches to find needed information, using such sources as the Internet.
99CI 97–100 · exposure 100 · augmentation 100 · importance 3.7/5 · click for rater detail
Conduct searches to find needed information, using such sources as the Internet.
99| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Administrative and knowledge-work sectors have rapidly adopted AI search and LLM assistants for information retrieval, with widespread integration into office workflows, demonstrating fast, deep adoption already in production. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | AI search assistants are widely and rapidly adopted in office/administrative and professional services settings, though full displacement of the broader secretarial role is slower. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI search tools substantially augment human secretaries by retrieving information faster and from more sources than manual search, allowing workers to focus on synthesis and decision-making while remaining in control of query refinement and result evaluation. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI dramatically speeds up and improves information-gathering research for administrative staff, who remain in the loop to verify and apply results. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Current AI systems (search engines, LLMs, agents) can conduct comprehensive information searches across the internet end-to-end with minimal human intervention, meeting the 50% time-saving threshold reliably through tools like ChatGPT with browsing, Perplexity, and specialized search agents. |
| Task automatability | claude-sonnet-5 | 5/5 | Web/information search and synthesis is a core strength of current AI systems with browsing/tool use, easily meeting the 50% time-saving bar for most routine information-gathering tasks. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | No licensing, regulatory, or legal requirement mandates human involvement in searching public internet sources; no liability asymmetry specific to the search task itself; organizational adoption faces minimal friction. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or regulatory requirement mandates a human perform general internet searches; this is a low-stakes, unrestricted task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The marginal cost of AI-driven web search is negligible (API calls cost pennies per query) compared to human labor time, especially when accounting for overhead; an AI agent can search orders of magnitude faster and cheaper than a secretary. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI search/retrieval costs pennies per query versus a human employee's loaded hourly wage, representing well over an order-of-magnitude cost advantage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed products demonstrably perform this task reliably in production: search engines, LLM-based assistants with web access, and enterprise search tools have been in real-world use for years and handle general information retrieval at scale with high accuracy. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Deployed products like ChatGPT with browsing, Copilot, Perplexity, and Gemini reliably perform web research and information retrieval at scale in production today. |
Make copies of correspondence or other printed material.
97CI 95–100 · exposure 100 · augmentation 38 · importance 3.7/5 · click for rater detail
Make copies of correspondence or other printed material.
97| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Document automation and self-service copying have been deeply adopted across information-intensive sectors (corporate offices, legal firms, healthcare, education, finance) for decades, with high penetration. |
| Sector adoption velocity | claude-sonnet-5 | 5/5 | Copying and scanning automation has been standard practice across nearly all sectors for decades, representing essentially complete adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | While AI-driven document management assists humans by enabling faster retrieval and organization of copied materials, the core copying function itself offers limited augmentation—it is primarily a replacement task. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Since the task is already almost entirely automated, there is little role left for AI to augment a human performing it manually. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | This task is fully automatable: scanning/OCR and digital duplication are routine capabilities of current AI systems and document management software, achieving near-perfect quality with trivial time investment. |
| Task automatability | claude-sonnet-5 | 5/5 | Physical or digital copying is a fully mechanized, repetitive task already handled by copiers, scanners, and document management software with no quality loss. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal legal barriers exist; however, some organizational friction remains due to legacy workflows, need for physical document routing in certain contexts, and user preference for human-verified copies in sensitive contexts. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or human-contact requirements restrict copying tasks; it is already fully delegated to machines in most workplaces. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of automated copying (hardware amortized, electricity, minimal oversight) is orders of magnitude cheaper than paying a human wage to manually copy documents. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Machine copying costs fractions of a cent per page versus the loaded cost of a human manually duplicating documents, an order-of-magnitude or greater savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, widely deployed products (multifunction copiers, document management systems, scanning software) reliably perform this task at scale in production environments across organizations globally. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Photocopiers, multifunction printers, and cloud document tools reliably perform copying/duplication at scale in virtually every office today. |
Schedule and confirm appointments for clients, customers, or supervisors.
91CI 84–97 · exposure 92 · augmentation 88 · importance 3.9/5 · click for rater detail
Schedule and confirm appointments for clients, customers, or supervisors.
91| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Professional services, healthcare, education, and corporate sectors show rapid, measurable adoption of scheduling automation and AI-powered calendar assistants, with pilots and production deployments common. This reflects high digitization and information-sector prevalence of the task. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Scheduling automation is broadly adopted across offices, clinics, and service businesses, with mainstream calendar tools embedding AI scheduling features as standard, though very small or informal offices may lag. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI scheduling assistants significantly augment human administrative staff by automating routine back-and-forth, freeing them to handle exceptions, complex multi-party coordination, and relationship-building—transforming overall scheduling productivity while keeping humans in the loop for judgment calls. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI scheduling assistants substantially reduce time spent on coordination, suggest optimal slots, send reminders, and handle rescheduling, while humans retain oversight for exceptions or VIP clients. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Scheduling and confirming appointments is largely rule-based and can be substantially automated by current AI systems (calendar APIs, email/SMS confirmation tools, and scheduling assistants like Calendly integrated with AI agents). While edge cases around complex multi-party logistics exist, AI can handle the majority of routine scheduling with >50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 5/5 | Scheduling and confirming appointments is a well-structured, rule-based task involving calendar checks, availability matching, and templated communication—fully achievable via AI scheduling assistants and calendar-integrated agents today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist for appointment scheduling automation; organizations must maintain oversight for customer preference and data privacy, but nothing legally requires a human to perform or sign off on the scheduling itself. Adoption friction is mainly organizational inertia and preference for human contact in certain contexts. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, regulatory, or liability barriers exist for automated appointment scheduling; it is already widely delegated to software without legal or professional restriction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of AI-driven scheduling tools (often <$10–30/month per user or per API call) is substantially cheaper than the loaded hourly wage of an administrative assistant ($25–40/hour all-in), making automation at least an order of magnitude cheaper when amortized across appointments. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated scheduling tools cost a few dollars per month or per-seat licensing versus the substantial hourly wage of a human assistant performing the same repetitive task, yielding order-of-magnitude savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple mature products (Calendly, AI assistants integrated into Outlook/Google Calendar, scheduling bots via Slack/Teams) are deployed in production at scale and perform this task reliably, with real organizations using them to manage appointment scheduling and confirmations daily. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Products like Calendly, Microsoft Bookings, Google Calendar AI features, and dedicated scheduling assistants (e.g., x.ai, Reclaim.ai) are deployed at scale in production and reliably handle scheduling and confirmation workflows. |
Review work done by others to check for correct spelling and grammar, ensure that company format policies are followed, and recommend revisions.
91CI 84–97 · exposure 92 · augmentation 100 · importance 3.7/5 · click for rater detail
Review work done by others to check for correct spelling and grammar, ensure that company format policies are followed, and recommend revisions.
91| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Grammar and spell-check tools are already ubiquitous in office software and have achieved deep penetration across information work sectors; firms routinely deploy automated checking before human review, representing fast, measurable adoption. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative environments have rapidly adopted AI writing and editing tools embedded in common productivity suites (Word, Google Docs, Grammarly) at scale. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI tools like Grammaly and Microsoft Editor actively assist users by highlighting errors in real-time and suggesting revisions, dramatically raising the productivity of humans who remain in the loop to make final judgment calls on style and tone. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools substantially speed up and improve the quality of proofreading and format-checking while the secretary retains final judgment on revisions and company-specific policy nuances. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can reliably detect spelling and grammar errors and flag deviations from formatting standards using current language models and rule-based systems. While some edge cases involving context-dependent style choices may require human judgment, the core task of checking spelling, grammar, and basic format compliance can be automated to achieve >50% time savings with equal or better quality. |
| Task automatability | claude-sonnet-5 | 5/5 | Grammar/spelling checking and format compliance review is well within current AI capability, with tools like Grammarly and LLMs performing this at or above human accuracy and much faster. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or licensing barriers exist for automating grammar and format checking. The main friction is organizational preference for human review and potential liability concerns about missed errors, but nothing legally requires a human to perform this task. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, regulatory, or liability barriers prevent AI from checking spelling, grammar, and formatting; this is low-stakes clerical work. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered grammar and spell-checking tools cost pennies per document or are embedded in free/cheap software, while human review time commands loaded hourly wages; the cost difference is orders of magnitude in favor of automation. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated proofreading and format-checking tools cost a few dollars per user per month versus the loaded cost of human review time, an order of magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple production systems (Grammarly, Microsoft Editor, Google Docs review tools) are widely deployed and demonstrably perform spelling/grammar checking and basic style enforcement at scale in real organizations. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Grammarly, MS Editor, Google Docs suggestions, and LLM-based proofreading are mature, widely deployed products used at scale in real organizations today. |
Use computers for various applications, such as database management or word processing.
90CI 80–100 · exposure 87 · augmentation 100 · importance 4.2/5 · click for rater detail
Use computers for various applications, such as database management or word processing.
90| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Information and professional services sectors are rapidly deploying AI-powered office automation, with widespread production adoption of copilots and workflow automation tools in enterprise environments. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative software adoption of AI features is proceeding quickly given the broad enterprise deployment of AI-enhanced productivity suites across many sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI significantly augments human productivity on these tasks through real-time suggestions, auto-completion, error correction, and intelligent data organization, allowing humans to focus on higher-level judgment while remaining in the loop. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially boosts productivity in drafting, editing, formatting, and data organization tasks while the human remains in control of final outputs. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Database management and word processing are fully automatable end-to-end with current AI systems using agents and RPA tools, delivering well over 50% time savings through automated data entry, document generation, formatting, and query execution. |
| Task automatability | claude-sonnet-5 | 4/5 | Word processing and database tasks like drafting, formatting, data entry, and simple queries are largely automatable with current AI tools including Copilot-style assistants and agentic workflows, though some tasks still require human setup or judgment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | There are no legal, licensing, or regulatory barriers preventing AI automation of generic database management and word processing; these are routine administrative tasks with no human-contact or signature requirements. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There are no licensing, legal, or regulatory requirements around using computers for word processing or database management; adoption is purely an organizational choice. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference and integration costs for database and word processing tasks are orders of magnitude cheaper than loaded human wages, especially for repetitive operations like form filling, data entry, and document formatting. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted document and data tools cost a small fraction of a secretary's hourly wage per unit of output, though integration and review overhead reduce the savings somewhat. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, deployed products (Microsoft Copilot, Claude, specialized RPA platforms) reliably perform word processing and database tasks in production environments across thousands of organizations at scale. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products (Microsoft 365 Copilot, Google Workspace AI, database automation tools) are in wide production use for drafting documents and managing structured data, though accuracy on complex database logic still requires oversight. |
Maintain scheduling and event calendars.
86CI 75–97 · exposure 87 · augmentation 88 · importance 3.8/5 · click for rater detail
Maintain scheduling and event calendars.
86| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Calendar automation is already widely adopted in information and professional services sectors; most modern organizations use some form of automated scheduling. Adoption is rapid in digital-first companies, though slower in small firms or highly regulated industries, placing it in the upper adoption category. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and office support functions in professional services and corporate settings have rapidly adopted AI scheduling tools, with widespread integration into email and calendar platforms. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists secretaries by auto-scheduling, conflict detection, and meeting prep (pulling attendee details, room bookings), freeing human attention for higher-value coordination, follow-up, and exception handling. The human remains in the loop for policy decisions, but productivity gains are substantial. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI scheduling assistants substantially reduce the back-and-forth and cognitive load of calendar coordination, letting administrative staff focus on more complex tasks while AI handles routine scheduling. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Calendar management is highly structured and repetitive: receiving meeting requests, parsing dates/times, identifying conflicts, and updating calendar systems. Current AI tools (Outlook/Google Calendar automation, scheduling assistants like Calendly) can fully automate this task with well over 50% time savings at equal or better quality, requiring only email integration or calendar API access. |
| Task automatability | claude-sonnet-5 | 4/5 | Calendar management is highly structured and rule-based, and AI scheduling assistants can already handle booking, rescheduling, and conflict resolution with minimal human input for most routine cases. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist for calendar automation. The main friction is organizational preference (some executives want personal assistant touch), data sensitivity in some industries, and integration setup. No license or human sign-off requirement applies to scheduling itself. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There are no licensing, legal, or regulatory requirements around calendar management, and no inherent need for human judgment or authorization. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of calendar automation (API calls, subscription tools like Calendly or Microsoft Bookings) is minimal—typically $0.001–$10 per month per calendar at scale—versus a loaded hourly wage for a secretary ($20–$35/hour) spending time on scheduling. AI cost is orders of magnitude lower. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated scheduling tools cost a small fraction of a human assistant's hourly wage for the same volume of calendar management, though some integration and oversight costs remain. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed calendar automation products are mature and widely used in production across organizations of all sizes. Scheduling assistants, email-integrated calendar tools, and calendar parsing via AI are reliable, with error rates low enough for routine deployment without human oversight in most contexts. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like AI scheduling assistants, Microsoft Copilot, and Google Calendar's AI features are deployed at scale and reliably handle meeting scheduling, though edge cases (VIP preferences, complex multi-party negotiations) still need human oversight. |
Compose, type, and distribute meeting notes, routine correspondence, or reports, such as presentations or expense, statistical, or monthly reports.
86CI 79–92 · exposure 83 · augmentation 100 · importance 3.8/5 · click for rater detail
Compose, type, and distribute meeting notes, routine correspondence, or reports, such as presentations or expense, statistical, or monthly reports.
86| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Professional services, finance, and information-sector firms have rapidly adopted AI tools for document generation and administrative automation; pilot projects and production deployments of AI-assisted writing are widespread in digitized sectors with white-collar administrative functions. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Office/administrative functions across many sectors are rapidly adopting AI writing assistants embedded in common productivity software, though full-scale replacement of drafting tasks is still uneven across smaller firms. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI significantly augments administrative staff by drafting documents, organizing notes, and generating first drafts that humans refine and approve, substantially raising per-person output while maintaining human oversight and quality control. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially speeds up drafting, formatting, and summarizing while the secretary remains in the loop for review, distribution, and personalization, making this a strong augmentation case. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI can reliably compose routine meeting notes, correspondence, and standard reports from source materials or transcripts with minimal human intervention, meeting the 50% time-saving threshold. The task involves templated, lower-complexity writing that large language models handle well, though highly context-specific or sensitive correspondence may still need review. |
| Task automatability | claude-sonnet-5 | 5/5 | Drafting meeting notes, routine correspondence, and standard reports from inputs/transcripts is well within current LLM capability, often exceeding the 50% time-saving threshold with tools like Copilot or Gemini integrated into office suites.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard regulatory or legal barriers exist for composing routine administrative documents; however, organizational friction (preference for human review, liability concerns, change management) and the need for human sign-off on sensitive correspondence impose moderate friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, legal sign-off, or human-contact requirement governs drafting internal notes or routine reports; organizations can adopt AI drafting freely. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The inference cost of generating standardized documents is negligible (fractions of a cent per task), while a human administrative assistant costs $25–50+ per hour loaded; AI is easily an order of magnitude cheaper even accounting for oversight and integration overhead. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI subscription costs per document are cents to a few dollars versus a loaded administrative wage per hour, making AI drafting an order of magnitude cheaper for this repetitive text task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (e.g., GPT-4, Claude, specialized business tools) demonstrably generate meeting notes, expense reports, and standard correspondence in production settings with acceptable quality for routine use. Minor factual errors or tone mismatches occasionally require human review, but the technology is mature enough for real-world deployment. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products (Microsoft 365 Copilot, Google Workspace AI, Otter.ai) already generate meeting summaries and draft correspondence in production, though human review is still typical for polish and accuracy. |
Operate electronic mail systems and coordinate the flow of information, internally or with other organizations.
83CI 75–91 · exposure 80 · augmentation 100 · importance 4.0/5 · click for rater detail
Operate electronic mail systems and coordinate the flow of information, internally or with other organizations.
83| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Email automation and AI-assisted inbox management are widespread in corporate, government, and professional-services settings; auto-responders, intelligent filtering, and scheduling are standard, reflecting rapid and deep adoption across digitized sectors. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and office support functions across many sectors have rapidly adopted AI email and scheduling assistants, reflecting fast adoption patterns typical of information-work tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI tools demonstrably increase human productivity by auto-drafting responses, highlighting urgent messages, organizing folders, and scheduling—transforming the secretary's ability to manage email volume while they focus on judgment-heavy coordination tasks. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially boosts productivity in this task today—auto-drafting replies, prioritizing inboxes, and summarizing threads—while humans retain final judgment and oversight. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Email operations and basic information routing are highly amenable to automation; AI agents can filter, categorize, draft responses, schedule distribution, and manage filing with minimal human intervention. However, judgment on sensitive or ambiguous messages and stakeholder coordination may still require human oversight, preventing a full 5. |
| Task automatability | claude-sonnet-5 | 4/5 | AI email tools can draft, sort, summarize, route, and even respond to routine correspondence with substantial time savings, though full end-to-end coordination across systems still often needs human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or licensing barriers exist for email automation; the main friction is organizational (need for integration, user adoption, oversight of sensitive messages) and customer preference to interact with human receptionists, but these are soft rather than hard barriers. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or legal requirement mandates a human to manage email flow; organizations freely adopt automated tools for this without regulatory constraint. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Email automation infrastructure costs fractions of a cent per message once deployed; a single AI-powered email system serves hundreds of users, making per-task cost orders of magnitude below the loaded human wage for equivalent output. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-assisted email management tools are inexpensive relative to a secretary's loaded wage, often bundled into existing software subscriptions, making per-task cost far lower than human labor for high-volume routine handling. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature email management systems with AI-powered sorting, spam filtering, auto-reply, and workflow automation are deployed at scale across enterprises today; tools like Microsoft 365 and Google Workspace integrate these capabilities into production systems millions use daily. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products (Outlook/Gmail AI features, Copilot, scheduling assistants) reliably handle email triage, drafting, and summarization in production today, though complex multi-party coordination is less mature. |
Mail newsletters, promotional material, or other information.
83CI 75–91 · exposure 80 · augmentation 75 · importance 3.4/5 · click for rater detail
Mail newsletters, promotional material, or other information.
83| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Bulk mailing automation has been widely adopted across marketing, administrative, and corporate communications for years. Email marketing and postal services routinely deploy these automations in production, reflecting mature, deep penetration in information-sector and professional-services organizations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and marketing functions have rapidly adopted automated mailing/email tools across most office-based sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI and automation tools substantially assist humans in list segmentation, content personalization, scheduling optimization, and delivery tracking, raising administrative productivity without removing human oversight of messaging and targeting decisions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI assists significantly with content drafting, personalization, and distribution logistics while a human still oversees strategy and final approval. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | The task of mailing newsletters, promotional materials, or information is highly automatable through existing mail-merge, distribution list management, and postal/email automation tools. Current systems can handle address verification, formatting, postage calculation, and batch sending with minimal human intervention, easily achieving 50%+ time savings. |
| Task automatability | claude-sonnet-5 | 4/5 | Physical mailing logistics (address list management, mail merge, scheduling bulk sends) can largely be automated via existing software, though physical mail insertion/stuffing may need human or mechanical handling. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers prevent automation of this task. Minor friction exists around customer preference for personalized touch and list management compliance (GDPR/CAN-SPAM), but these do not require human signature-off and are easily managed through existing tooling. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or regulatory barriers prevent automating newsletter or promotional mailings. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated mailing via email or postal service providers costs orders of magnitude less per piece than manual labor once fixed setup is amortized, especially for digital distribution. A human would require significant time for addressing, sorting, and postage handling; automation reduces marginal cost to near-zero for email and pennies per piece for physical mail. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated mailing/email platforms cost far less per unit than manual staff time for addressing and dispatching materials. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, production-grade systems (email marketing platforms, mail service integrations, CRM tools) reliably perform bulk mailing at scale across millions of organizations. Services like Mailchimp, Constant Contact, and postal automation providers demonstrate reliable, repeatable performance in real deployments. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mail merge tools, email marketing platforms, and bulk mail services are mature, widely deployed products handling this reliably at scale today. |
Complete forms in accordance with company procedures.
82CI 72–92 · exposure 87 · augmentation 75 · importance 3.8/5 · click for rater detail
Complete forms in accordance with company procedures.
82| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Administrative and back-office automation is rapidly deployed across finance, HR, and professional services sectors. RPA and intelligent process automation adoption is accelerating in digitized organizations, though smaller firms lag. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative/clerical functions are seeing moderate automation adoption via RPA and AI form tools, but many organizations still rely on manual processes for exception handling. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can draft, pre-fill, and validate forms in real-time, dramatically raising human productivity when humans remain responsible for review and approval. AI assistance allows staff to handle higher volumes and focus on judgment-based tasks. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools substantially speed up form completion by auto-filling fields, extracting data from source documents, and flagging errors, while humans still verify accuracy and handle exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Form completion is highly structured data entry following defined rules and procedures. Current AI systems (including document processing agents with OCR, table extraction, and rule-based logic) can consistently fill standardized forms by 50%+ time savings while matching human quality, especially with pre-populated templates. |
| Task automatability | claude-sonnet-5 | 4/5 | Form-filling is a structured, rule-based task well-suited to AI/automation tools that can extract data and populate fields, though some forms require judgment or unstructured input handling that needs setup. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Most form completion has low legal/regulatory barriers; however, some administrative contexts (compliance, regulated industries) may require human sign-off, and organizational processes often embed human approval steps. No licensing requirement for the automation itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Some forms require human verification or signature depending on company policy, but there is generally no legal/licensing requirement for a human to complete administrative forms. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | API-based form automation costs pennies per form processed; typical administrative assistant hourly cost is $25–40 per hour. For routine forms, AI cost is at least 10–100× cheaper per completed form including integration overhead. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated form completion via RPA/AI is typically far cheaper per form than manual entry once set up, though initial integration costs reduce the ratio somewhat versus pure inference cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature products (Zapier, RPA platforms, intelligent document processing systems, GPT-4-based form-filling agents) reliably complete routine forms in production workflows at scale across finance, HR, and administrative operations. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | RPA tools, document AI, and workflow automation products (e.g., form-fill bots, DocuSign integrations, OCR-based systems) are widely deployed in production for standardized business forms today. |
Prepare conference or event materials, such as flyers or invitations.
82CI 80–84 · exposure 75 · augmentation 100 · importance 3.3/5 · click for rater detail
Prepare conference or event materials, such as flyers or invitations.
82| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Marketing, events, and corporate communications departments are rapidly adopting AI design tools in production workflows; Canva and similar platforms report widespread SMB and enterprise use for exactly this task. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | AI-assisted design and content tools have seen fast, broad adoption across office and administrative settings, with many secretaries already using such tools for routine materials. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically assists humans by generating draft layouts, color schemes, and copy instantly, allowing administrators to focus on brand alignment and messaging—transforming productivity while human oversight remains. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools dramatically speed up and improve the quality of flyer/invitation creation while the secretary retains control over content, branding, and final approval. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can generate layouts, text, and graphics for flyers and invitations with high quality using tools like DALL-E, GPT, and design platforms; minimal human input is needed beyond specifying tone and key details, easily achieving 50% time savings. |
| Task automatability | claude-sonnet-5 | 4/5 | Drafting flyers, invitations, and event materials from templates or prompts is a well-suited generative AI task; design tools with AI (Canva, Microsoft Designer) can produce polished drafts in minutes, saving significant time though final selection/editing still needed. |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | No licensing, regulatory, or legal requirements constrain automation of flyer and invitation design; organizations face no liability asymmetry and internal customers typically accept AI-generated materials. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, regulatory, or liability barriers exist for creating flyers or invitations; organizations can freely adopt AI tools for this purpose. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered design tools cost pennies per asset generated, with minimal human oversight; this is orders of magnitude cheaper than paying a human designer or administrative assistant for equivalent output. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI design tools cost a small subscription fee versus staff time to design materials manually, making AI substantially cheaper per output, though some human review/editing time remains. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple production-grade systems (Canva with AI, Adobe Express, DALL-E integration, and template-based design tools) reliably generate event materials at scale; while some refinement may be needed, the core task is demonstrably performable in deployed products. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like Canva AI, Adobe Express, and Microsoft Designer are widely deployed and reliably generate professional-looking flyers and invitations today with minimal user input. |
Perform payroll functions, such as maintaining timekeeping information and processing and submitting payroll.
80CI 70–90 · exposure 87 · augmentation 75 · importance 4.2/5 · click for rater detail
Perform payroll functions, such as maintaining timekeeping information and processing and submitting payroll.
80| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Payroll automation is among the earliest-adopted enterprise automation use cases, with decades of maturity and near-universal adoption in medium and large firms. Most organizations have already migrated away from manual payroll processing to automated systems. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Payroll software adoption is widespread and mature across nearly all industries and firm sizes, representing one of the most digitized back-office HR functions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered payroll systems augment human administrators by providing real-time dashboards, anomaly detection (unusual hours, missing data), automated compliance alerts, and self-service employee portals, allowing humans to focus on exceptions and policy decisions rather than routine data entry and calculation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-enabled payroll platforms significantly reduce manual entry and error-checking burden, letting administrative assistants focus on exceptions, corrections, and employee inquiries. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Payroll processing is highly structured, rule-based work with clear inputs (timekeeping data, salary rates, tax tables) and outputs (payroll records, disbursements). Current AI and RPA systems can handle end-to-end payroll workflows—data entry, calculation, compliance checks, and submission—with significant time savings and equal or better accuracy than manual processing. |
| Task automatability | claude-sonnet-5 | 4/5 | Timekeeping data entry and payroll processing follow structured, rule-based workflows that current payroll software and AI-augmented systems (e.g., Gusto, ADP, Workday) can largely automate, though exception handling and approvals still need human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Payroll is regulated (tax law, labor law, wage-hour compliance), and human review/sign-off is often required for accuracy and liability reasons. However, these are oversight requirements rather than hard legal prohibitions on automation, and many organizations have already implemented automated payroll systems with human supervisory controls. |
| Adoption barriers | claude-sonnet-5 | 3/5 | While no license is legally required to run payroll, error-cost asymmetry (tax penalties, wage law violations) and organizational need for accountable sign-off create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Payroll software costs per employee per month are typically $5–$15 all-in, compared to a loaded secretary wage of $35–$55/hour for payroll tasks that could take several hours monthly. AI/software is at least an order of magnitude cheaper when amortized across employees and transactions. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated payroll systems process large volumes of transactions at a fraction of the marginal labor cost per employee-pay-cycle, though initial setup, integration, and oversight costs temper full order-of-magnitude savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature payroll automation products (ADP, Gusto, Workday, BambooHR) are deployed at scale in production across organizations globally. These systems reliably handle timekeeping integration, tax calculations, direct deposit, and regulatory compliance with minimal error rates and established audit trails. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mature payroll software products already automate most of timekeeping aggregation, tax calculations, and submission in production at scale for many organizations, though edge cases and compliance checks still involve human review. |
Provide services to customers, such as order placement or account information.
80CI 79–81 · exposure 75 · augmentation 75 · importance 3.8/5 · click for rater detail
Provide services to customers, such as order placement or account information.
80| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Adoption of AI-driven customer service is already widespread in retail, banking, and telecom sectors, with major companies operating mature chatbot and IVR systems in production. This is among the fastest-adopting task categories across occupations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Customer service functions across e-commerce, telecom, and finance have seen fast, deep AI adoption with chatbots and virtual assistants now standard in production. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI augmentation in customer service is strong: AI can draft responses, pull account data, flag escalation triggers, and suggest next steps, significantly raising a human agent's throughput and accuracy when handling complex or sensitive inquiries. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up secretaries' handling of routine customer requests by pulling account data, drafting responses, and pre-filling order details, letting humans focus on exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Order placement and account information retrieval are highly structured tasks with well-defined workflows that current AI systems handle via chatbots and automated support agents. While some complex account issues may still require human judgment, the majority of routine customer service interactions can be fully automated, meeting the ≥50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 4/5 | Order placement and account information retrieval are highly structured, rules-based tasks that chatbots and voice AI already handle end-to-end for a large share of routine interactions, though escalations and edge cases still need humans. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or licensing barriers prevent automation of routine customer service. The main friction points are organizational preference for human touch and customer experience concerns, neither of which are hard legal barriers to substitution. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for this task, though some organizational friction and customer preference for human contact in complex cases creates mild resistance. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Conversational AI inference cost per interaction is typically cents, while a loaded secretary wage for the same task runs $15–30 per hour or more. The cost differential favors AI by a clear order of magnitude when integrated at scale. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated order/account systems cost a fraction of a cent to a few cents per interaction versus a loaded human wage, giving at least an order-of-magnitude cost advantage for routine transactions. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (chatbots, automated phone systems, live chat agents with AI) demonstrably handle order placement and account lookups in production across retail, finance, and service sectors. Performance is generally reliable for standard inquiries, though edge cases and context-dependent issues still occur. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed AI customer service agents (e.g., chatbots, IVR systems, CRM-integrated assistants) handle order placement and account inquiries reliably at scale across retail, telecom, and banking today, though not universally perfect. |
Arrange conference, meeting, or travel reservations for office personnel.
79CI 75–84 · exposure 75 · augmentation 88 · importance 3.7/5 · click for rater detail
Arrange conference, meeting, or travel reservations for office personnel.
79| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Large corporations and professional-services firms have already deployed calendar and travel AI extensively; mid-market firms are adopting rapidly. This is a mature automation pattern in digitized, information-sector workplaces where the task occurs. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Administrative and office support functions across many industries have rapidly adopted calendar and travel-booking automation tools, reflecting fast adoption typical of digitized office workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistants that surface availability, suggest optimal travel options, flag conflicts, and auto-populate forms significantly boost human productivity on this task, even when humans retain final approval authority. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI significantly boosts productivity for this task by auto-suggesting times, handling back-and-forth coordination, and pre-filling travel details, while the human retains final decision-making for exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can now handle most of the core work—searching for dates, comparing prices, booking flights/hotels, and managing calendar conflicts—with minimal human intervention. End-to-end automation of standard reservations easily clears the 50% time-saving bar; only edge cases (unusual preferences, complex multi-leg coordination, last-minute changes) require human oversight. |
| Task automatability | claude-sonnet-5 | 4/5 | Scheduling meetings, booking travel, and coordinating logistics are well-structured tasks that AI scheduling assistants and travel booking agents can already handle with minimal human intervention, though occasional edge cases (visa issues, complex multi-party conflicts) still need human judgment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard regulatory or licensing barriers exist for automated travel booking. Corporate travel policies and preference for human review on sensitive trips introduce some friction, but nothing legally mandates human involvement in routine reservation-making. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There is no licensing, regulatory, or liability barrier to using AI for scheduling meetings or booking travel; organizations widely permit software tools to handle this already. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference and integration cost for booking tasks is negligible—typically cents per reservation—compared to the fully-loaded wage of a human administrative assistant (often $30–50K annually or $15–25/hour). Even accounting for oversight and exception handling, AI is an order of magnitude cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI scheduling and booking tools operate at a fraction of the cost of a human assistant's time for these routine coordination tasks, though integration and occasional human correction add some overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed products (calendar AI, travel booking agents, Slack/Teams integrations) perform this at scale in enterprise settings. Error rates on routine bookings are low; integration with corporate travel policies and approval workflows is mature, though some friction remains around policy exceptions. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products like AI calendar assistants (e.g., Clara, Reclaim, Microsoft Copilot) and corporate travel platforms (e.g., TripActions/Navan with AI booking) reliably perform scheduling and travel arrangement in production today, though some manual oversight remains common. |
Order and dispense supplies.
76CI 72–80 · exposure 75 · augmentation 75 · importance 3.3/5 · click for rater detail
Order and dispense supplies.
76| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Supply chain and inventory automation is widespread in medium-to-large organizations and has been adopted rapidly across finance, healthcare, manufacturing, and corporate sectors. Smaller firms lag, but overall adoption in digitized workforces is substantial and accelerating. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | While administrative/office functions are increasingly digitized, many small and mid-sized organizations still rely on manual processes for supply ordering, giving moderate but uneven adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered inventory systems assist secretaries by flagging low stock, recommending suppliers, and automating routine reorders, freeing them to focus on exception handling and relationship management. Real-time dashboards and predictive alerts significantly raise human productivity on this task. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered inventory trackers and reorder alerts substantially reduce the manual effort of monitoring stock and placing orders, letting assistants focus on exceptions and physical dispensing. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Inventory tracking, reordering triggers, and supply distribution can be largely automated through warehouse management systems and automated fulfillment workflows. While physical dispensing may require human handling in some contexts, ordering logic and distribution workflows can achieve >50% time savings with current systems. |
| Task automatability | claude-sonnet-5 | 4/5 | Ordering and dispensing supplies is largely a rule-based, transactional workflow (tracking inventory, reordering thresholds, purchase orders) that AI-driven procurement/inventory systems can handle with minimal human input, though physical dispensing still requires a person. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers exist for automating supply ordering and dispensing; no licensed professional is required to sign off. Main friction points are organizational inertia and the preference for human touch in some workplace cultures, not regulatory constraints. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There are no licensing, legal, or safety requirements around ordering office supplies; it's a low-stakes administrative task with no regulatory protection. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automation of ordering and inventory management is significantly cheaper than manual oversight once systems are deployed. Software licensing and integration costs are low relative to the labor hours saved in tracking, ordering, and distribution workflows. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated procurement/inventory software is inexpensive relative to paying a human to manually track and order supplies, though some human oversight and physical handling costs remain. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products exist for inventory management, automated reordering, and supply chain coordination (e.g., SAP, Coupa, modern ERP systems). These systems reliably perform ordering and dispatch functions at scale in production environments, though final physical dispensing may still involve human workers. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Automated inventory management and procurement software (e.g., auto-reordering systems integrated with vendor platforms) are widely deployed in offices today, though the physical distribution of supplies still needs human action. |
Set up and manage paper or electronic filing systems, recording information, updating paperwork, or maintaining documents, such as attendance records, correspondence, or other material.
76CI 72–79 · exposure 75 · augmentation 75 · importance 4.0/5 · click for rater detail
Set up and manage paper or electronic filing systems, recording information, updating paperwork, or maintaining documents, such as attendance records, correspondence, or other material.
76| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Document management and RPA adoption is accelerating across corporate and administrative sectors, particularly in information-heavy industries (finance, professional services, healthcare administration). Many organizations have already deployed systems or are in active pilots. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and clerical functions in offices are moderately digitized, with many small and mid-sized organizations still relying on manual or hybrid filing methods despite pilots of automated document systems. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists humans significantly by auto-categorizing documents, flagging updates needed, suggesting filing locations, and maintaining audit trails—allowing administrative staff to focus on exception handling and more strategic organizational tasks rather than manual filing. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up filing, indexing, searching, and updating documents, letting administrative assistants manage larger volumes of records more efficiently while still verifying accuracy and handling exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can handle much of the filing, categorization, and document organization work (OCR, classification, metadata extraction) and electronic record management nearly end-to-end, potentially saving >50% of time on routine filing and updates. However, some judgment about document priority, sensitive information handling, and organizational context still typically requires human oversight. |
| Task automatability | claude-sonnet-5 | 4/5 | Filing, organizing, and updating documents is largely structured information management that current AI (document management systems, automated tagging, OCR, cloud-based filing with search) can handle with significant time savings, though initial setup and edge-case handling still need human input. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Barriers are minimal: no licensing requirement, no legal mandate for human sign-off, and few organizational or liability constraints prevent automation. Some organizations prefer human document handling for sensitive materials, but this is preference-based rather than regulatory. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Minimal licensing or legal requirements apply to general filing tasks, though some organizations require compliance with retention or privacy policies (e.g., HR or personnel records) that impose light oversight. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven document management and RPA systems cost a fraction of human labor for filing and record-keeping tasks, often at least an order of magnitude cheaper per document processed when amortized across organizations. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Cloud storage and automated filing tools cost a small fraction of a human's hourly wage for routine filing and updating tasks, though occasional oversight and correction still incur some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed RPA platforms, document management systems, and AI-powered filing solutions perform this task reliably in production across many organizations. End-to-end automation of paper-to-electronic conversion and file maintenance is mature, though quality can vary with document complexity and custom categorization rules. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Mature products like SharePoint, Google Workspace, DMS platforms with AI-based auto-categorization and metadata tagging are widely deployed and reliably used in production for document/records management today. |
Prepare and mail checks.
74CI 70–79 · exposure 75 · augmentation 63 · importance 3.7/5 · click for rater detail
Prepare and mail checks.
74| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Finance and accounting functions have strong adoption of RPA and payment automation; medium to large organizations routinely deploy these systems. Smaller firms and those with legacy systems lag, but adoption is measurably rapid in the sectors where this task is prevalent. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Financial and administrative back-office functions have seen fast, broad adoption of automated payment and accounts-payable software across firms of many sizes. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems can assist by pre-filling check data, validating addresses, flagging exceptions, and auto-scheduling mail, all while a human retains control over authorization and final dispatch. This substantially raises the productivity of secretaries who retain oversight. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Even where automated, humans often still review, approve, and reconcile payments, so AI meaningfully speeds the task while keeping oversight in place. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Check preparation and mailing is highly procedural: data entry, document generation, and physical logistics can be automated end-to-end with current RPA and mail-integration tools, easily achieving 50% time savings. The main remaining variable is authorization/signing, which may require human oversight depending on organizational policy. |
| Task automatability | claude-sonnet-5 | 4/5 | Preparing and mailing checks is largely rules-based data entry and workflow execution, which AI-integrated accounting/payment software can handle end-to-end with minimal human input beyond approval. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Financial controls and audit requirements typically mandate human review or signature on checks above certain thresholds, and some organizations retain manual oversight for fraud prevention. However, no universal legal licensing requirement prevents automation itself. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Some organizational friction exists around authorization and signatory controls, but no licensing requirement mandates a human perform this task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automating check preparation via RPA or accounting-system plugins costs pennies per transaction in infrastructure and oversight, vastly cheaper than the fully-loaded wage of a secretary performing this task manually. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated check-writing and mailing services cost a small monthly/per-transaction fee, far below the loaded cost of manual secretarial time for the same volume. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed RPA platforms and accounting software integrations already handle check generation, printing, and address validation at scale in many organizations. Mail services and payment APIs are mature, though full end-to-end automation still often requires human approval steps, limiting the 5 rating. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products like QuickBooks, Bill.com, and ERP payment modules reliably automate check preparation and even outsource physical mailing at scale in production today. |
Create, maintain, and enter information into databases.
74CI 72–75 · exposure 75 · augmentation 63 · importance 4.2/5 · click for rater detail
Create, maintain, and enter information into databases.
74| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Database and administrative automation are among the fastest-adopted use cases in business process automation. Finance, healthcare, retail, and government organizations have already deployed RPA and automation extensively for these tasks, showing strong production adoption and measured displacement. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support functions are adopting AI/RPA at a moderate pace with many pilots and partial deployments, but full-scale replacement in production remains inconsistent across firms of varying size. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI and automation provide useful assistance by auto-completing fields, flagging anomalies, and validating data quality, which can speed up human review and correction work. However, when fully automated, the human is removed from the loop rather than empowered to work alongside the system, so augmentation is partial rather than transformative. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools significantly speed up data entry, validation, and organization while a human still verifies quality control and handles exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Database creation, maintenance, and data entry are highly structured, rule-based tasks that current AI and automation tools can perform end-to-end. RPA (Robotic Process Automation), data integration platforms, and AI-driven ETL solutions can accomplish this with significant time savings and at least equal quality, easily meeting the 50% threshold for most standard database operations. |
| Task automatability | claude-sonnet-5 | 4/5 | Data entry and database maintenance are structured, repetitive tasks well within current AI/RPA capability, especially when paired with OCR and structured templates, though some tasks require judgment on ambiguous or unstructured source data. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal hard barriers exist for database automation—no licensing requirement mandates human involvement, and liability is typically contained via audit logs and data governance oversight. Some organizational friction and change management occur, but these are not structural legal or regulatory barriers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal sign-off requirements typically apply to database entry; main friction is organizational inertia and data quality/integration concerns rather than hard regulatory barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI and automation solutions for database work cost significantly less than human labor once deployed: no benefits, negligible per-transaction overhead, and continuous 24/7 operation. The all-in cost per entry or maintenance task is typically one or two orders of magnitude lower than a loaded human wage. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated data entry and database tools cost a fraction of a human's hourly wage per record processed, though initial integration and error-correction oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature, production-deployed products (including RPA platforms like UiPath and Blue Prism, database automation tools, and cloud-native data pipelines) reliably perform database creation, maintenance, and data entry at scale in real organizations today. Error rates are minimal for well-defined schema and data validation rules. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | RPA tools, AI-powered data entry systems, and integrations with CRM/ERP software are widely deployed in production today for creating and updating database records, though edge cases still require human review. |
Locate and attach appropriate files to incoming correspondence requiring replies.
72CI 67–77 · exposure 70 · augmentation 88 · importance 3.7/5 · click for rater detail
Locate and attach appropriate files to incoming correspondence requiring replies.
72| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Mid-market and enterprise organizations adopt these capabilities, but smaller firms and non-digitized sectors lag. Adoption is increasing but not yet ubiquitous; many organizations still perform this task manually despite available solutions. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative/office support functions are adopting AI tools steadily (email assistants, copilots), but many organizations still rely on manual file management, so adoption is moderate rather than fast and deep. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments this task by auto-retrieving candidate files and suggesting attachments, allowing the secretary to verify and send rather than manually searching. This dramatically accelerates the workflow while maintaining human oversight of file selection appropriateness. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI search and retrieval tools significantly speed up locating and attaching files for administrative staff, who remain in the loop to verify correctness before sending replies. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can reliably match incoming correspondence to relevant files using document classification, entity recognition, and contextual retrieval with high accuracy. This task is largely deterministic pattern-matching that email routing systems and document management systems already automate, saving substantial processing time. |
| Task automatability | claude-sonnet-5 | 4/5 | AI file search, retrieval-augmented systems, and email/document integrations can locate and attach relevant files with minimal human intervention once connected to organizational systems, saving substantial time. Full end-to-end reliability depends on integration quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Barriers are minimal: file attachment automation requires only organizational buy-in and IT setup, with no licensing or regulatory requirements. Some organizations may prefer human review for sensitive correspondence, but no legal mandate prevents automation. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, legal, or safety barriers exist; this is a low-stakes clerical task with no human-contact or liability requirement. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Cloud-based AI-powered document retrieval and email automation are extremely inexpensive per-task, typically costing pennies when amortized across an organization. This is orders of magnitude cheaper than paying a secretary's loaded wage to manually search and attach files. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Once integrated, AI-driven file retrieval and attachment costs a fraction of a cent per instance versus the loaded cost of a human assistant performing the same repetitive lookup task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Document management systems and email automation platforms with AI-powered file attachment and retrieval are in production use today. Major vendors (Microsoft 365, Google Workspace plugins, dedicated DMS platforms) offer reliable file-matching capabilities, though occasional manual review may be needed for ambiguous cases. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like Outlook/Gmail add-ins, enterprise search tools, and AI copilots (e.g., Microsoft 365 Copilot) can surface and attach related files today, but accuracy in messy or poorly indexed file systems remains inconsistent in production. |
Open, read, route, and distribute incoming mail or other materials and answer routine letters.
72CI 67–77 · exposure 70 · augmentation 75 · importance 3.7/5 · click for rater detail
Open, read, route, and distribute incoming mail or other materials and answer routine letters.
72| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Administrative functions see moderate adoption of document automation and RPA, with many large organizations piloting or deploying mail-sorting and response generation. However, uptake in small offices and non-digitized sectors remains slow, and full end-to-end replacement is less common than partial automation. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Office/administrative functions in many sectors have adopted email filters, chatbots, and routing tools, but comprehensive automation of mixed physical/digital mail workflows remains uneven and pilot-stage in many firms. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists by drafting responses, highlighting priority mail, and pre-sorting, enabling secretaries to review and refine output rather than start from scratch. This raises productivity modestly, though human judgment on sensitive or complex correspondence remains valuable. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools like smart inboxes, auto-categorization, and draft-response generators substantially speed up a secretary's handling and answering of routine correspondence while keeping them in the loop for judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Most of this task is directly automatable: mail can be scanned and digitized, content read by OCR/LLMs, routing rules applied via RPA, and routine responses generated by language models. The only friction is occasional physical mail handling and judgment calls on non-routine items, but the bulk workflow easily achieves >50% time savings at equal quality. |
| Task automatability | claude-sonnet-5 | 4/5 | Digital mail triage, classification, routing, and drafting routine responses are well within current LLM and workflow-automation capabilities, though physical mail opening/sorting still requires human or robotic handling.4. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few hard barriers exist: no licensing required, no legal mandate for human sign-off, and no strict regulatory restrictions on mail automation. Light friction remains around customer preference for human touch and internal organizational resistance, but these are soft, not legal or liability barriers. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, legal, or liability barriers prevent automating mail routing and routine correspondence; it's a low-stakes clerical function. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | OCR + routing automation + LLM-based responses cost pennies per item handled, versus a secretary's loaded wage of $25–40/hour. The AI cost per task equivalent is substantially lower once infrastructure is in place, typically 5–10x cheaper than human labor for high-volume mail. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated email triage and templated response systems cost a fraction of a cent per item versus a loaded clerical wage, though physical mail still needs human labor, moderating the overall ratio. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (enterprise document processing, email automation, RPA suites, and LLM-based response generation) already handle scanning, routing, and routine letter composition reliably in production. Some human review is typically retained, but the core technical capability is mature and in use at scale. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Email routing, ticketing, and auto-response tools are deployed widely, but physical mail handling and reliable end-to-end routing across varied document types still require human oversight in most organizations. |
Develop or maintain internal or external company Web sites.
69CI 55–84 · exposure 62 · augmentation 88 · importance 3.1/5 · click for rater detail
Develop or maintain internal or external company Web sites.
69| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Information and professional services sectors are rapidly adopting AI-assisted web tools and generative coding in production. Small and mid-market companies increasingly use no-code AI builders; adoption is visibly accelerating in digitized industries. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and clerical functions are adopting AI tools at a moderate pace—many small businesses use AI-assisted site builders, but full automation of ongoing web maintenance in production remains uneven across office settings. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI significantly augments human productivity in website maintenance through real-time code suggestions, content drafting, accessibility checking, and design alternatives, while the human retains strategic control and quality oversight. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI substantially speeds up drafting content, generating layouts, fixing code issues, and suggesting SEO improvements, making the secretary considerably more productive while still directing final decisions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems (generative code models, no-code tools) can automate substantial portions of website development and maintenance—content generation, basic layout design, code generation, and routine updates—achieving well over 50% time savings on equal quality. However, strategic direction and complex UX decisions typically require human judgment. |
| Task automatability | claude-sonnet-5 | 3/5 | AI website builders and code-generation tools can handle much of the design, content updates, and basic maintenance, but ongoing site management, integration with company systems, and troubleshooting still require human oversight for a non-technical secretary role. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Website development and maintenance carry minimal regulatory or licensing barriers; no legal sign-off is required. Barriers are primarily organizational inertia and preference for human oversight on brand-critical content, not structural legal constraints. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, regulatory, or liability barriers restrict who can build or maintain a company website; it's a purely administrative/technical task with no legal requirement for human sign-off. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference for code/content generation and maintenance costs pennies per task, versus a secretary's hourly wage (typically $30–50 loaded). The cost differential for routine website updates and maintenance is at least an order of magnitude in AI's favor. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-assisted website tools can reduce time spent on routine updates significantly, but licensing costs, integration effort, and the need for human review keep costs roughly comparable rather than dramatically cheaper for this often part-time task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products (GitHub Copilot, Claude Code, Webflow AI, CMS plugins with AI assistance) are deployed in production and reliably handle code generation, content updates, and design refinement. While some edge cases and brand-specific customization remain challenging, the core task is demonstrably performed by live systems. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like Wix ADI, WordPress AI plugins, and no-code builders with AI assistance are deployed and used broadly, but they still require human setup, review, and periodic intervention for reliability. |
Answer telephones and give information to callers, take messages, or transfer calls to appropriate individuals.
67CI 59–75 · exposure 62 · augmentation 63 · importance 4.3/5 · click for rater detail
Answer telephones and give information to callers, take messages, or transfer calls to appropriate individuals.
67| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Call automation is already deeply adopted in information, finance, healthcare, and customer service sectors; large enterprises deploy AI phone systems routinely, and smaller firms increasingly use cloud-based solutions, showing fast and measurable displacement of traditional reception roles. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative support functions are seeing growing AI adoption via virtual assistants and call-routing tools, but many small and mid-sized offices still rely on human staff for this task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist human receptionists by pre-screening callers, summarizing call context, and suggesting routing, enabling them to handle more complex calls efficiently; however, the task involves significant replacement rather than pure augmentation, limiting its augmentation-only upside. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI transcription, call summarization, and routing suggestions meaningfully boost efficiency for administrative assistants handling high call volumes while they remain in the loop for judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI call-handling systems can automate substantial portions—routing calls, capturing basic information, and transferring to appropriate departments—but many callers require live handling for nuanced questions or complex transfers, preventing full end-to-end automation with 50%+ time savings at equal quality for all call types. |
| Task automatability | claude-sonnet-5 | 4/5 | AI voice agents and IVR systems can handle call answering, information provision, message-taking, and routing with substantial time savings, though edge cases and nuanced routing still need human fallback., though not universally deployed. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or legal barriers exist to automating call answering; most barriers are organizational (customer preference for human contact, perceived quality concerns, change management friction) rather than compliance-based, and these preferences are eroding as systems improve. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for this task, but some organizational friction exists from customer preference for human contact and need for exception handling in ambiguous cases. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI call-handling systems cost a fraction of a full-time receptionist salary once deployed (typically $0.50–$2 per call in volume), making them 5–20× cheaper than human labor on a per-call basis, though integration and oversight add marginal costs. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Cloud-based AI phone assistants cost a fraction of a human receptionist's wage per call handled, though integration and monitoring add some overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed IVR and AI phone systems (from vendors like Amazon Connect, Google Cloud Contact Center AI, and others) reliably handle call answering, routing, and message-taking in production at scale across many organizations, though they still struggle with complex or atypical caller needs. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI phone answering products (e.g., virtual receptionists, conversational IVR) are deployed in production for many businesses, but accuracy and handling of complex or ambiguous requests remain limited, especially for less common scenarios. |
Train and assist staff with computer usage.
64CI 35–92 · exposure 58 · augmentation 88 · importance 3.4/5 · click for rater detail
Train and assist staff with computer usage.
64| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Information and professional services sectors, along with corporate training departments, have already adopted AI-driven training platforms, chatbots, and self-service help systems at measurable scale. Adoption is accelerating as organizations seek to reduce training overhead. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Administrative support roles are in a sector with slower AI integration for interpersonal training tasks, though general office AI tool adoption is rising slowly. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI can dramatically augment human trainers by generating personalized training materials, pre-answering common questions, providing real-time transcription and summaries, and creating follow-up resources—enabling each trainer to support many more staff members without leaving the role. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI chat assistants, tutorials, and documentation tools can significantly speed up how secretaries prepare training materials and answer common computer usage questions. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | AI systems can now create comprehensive training materials, video tutorials, step-by-step guides, and provide real-time interactive assistance for common software problems with high quality and consistency. End-to-end training delivery and staff support can achieve well over 50% time savings compared to human instruction, especially for routine, standardized software tasks. |
| Task automatability | claude-sonnet-5 | 2/5 | Training staff involves hands-on demonstration, adapting to individual skill levels, and answering ad hoc questions, which AI can support but not fully replace end-to-end in a live organizational setting.assign.5.setup |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | Training and assisting staff with computer usage has no legal licensing requirement, liability is low (errors are easily corrected and non-critical), and no human signature or approval is mandated. Adoption is driven entirely by preference and organizational friction, not regulatory gatekeeping. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human trainer, but organizational preference for personal rapport and immediate hands-on help creates moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The inference cost of AI-delivered training—combining automated video generation, chatbots, and knowledge base retrieval—is orders of magnitude cheaper than paying a staff member's loaded hourly wage to conduct one-on-one or group training sessions, even accounting for initial content setup and periodic updates. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI help tools are cheap per query, the human element of scheduling, tailoring, and hands-on troubleshooting means overall cost savings versus a secretary doing this task are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed AI products (ChatGPT, Claude, specialized helpdesk automation) successfully deliver training content, answer technical questions, and provide troubleshooting assistance in production environments across many organizations. While some edge cases and advanced technical issues require human intervention, the core task is reliably performable at scale today. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI chatbots and help documentation exist to assist with software questions, but reliable in-person or synchronous training and troubleshooting for staff is not yet handled by deployed products at scale. |
Collect and deposit money into accounts, disburse funds from cash accounts to pay bills or invoices, keep records of collections and disbursements, and ensure accounts are balanced.
62CI 50–74 · exposure 62 · augmentation 75 · importance 4.2/5 · click for rater detail
Collect and deposit money into accounts, disburse funds from cash accounts to pay bills or invoices, keep records of collections and disbursements, and ensure accounts are balanced.
62| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Finance and accounting functions in mid-to-large organizations have rapidly adopted RPA and intelligent accounting platforms; however, small firms and public-sector organizations lag, keeping overall velocity below the fastest-adopting sectors. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and small-business back-office functions show moderate AI/software adoption—automation tools are common but many organizations still keep humans in the loop for cash and reconciliation tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments human accountants and administrative staff by automating data entry, flagging discrepancies, and generating pre-reconciled reports, freeing them to focus on exception handling and analysis while remaining in the loop for approval and oversight. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered bookkeeping and reconciliation tools substantially speed up record-keeping and flag discrepancies, meaningfully boosting the productivity of the person managing these accounts. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can automate most of this task end-to-end: OCR/invoice processing captures bill data, robotic process automation (RPA) systems execute transfers between accounts, and reconciliation algorithms verify balances. However, some judgment (unusual transactions, exception handling) may require human review, preventing a full 5. |
| Task automatability | claude-sonnet-5 | 3/5 | Software can automate invoice processing, categorization, and reconciliation, but physical cash handling, bank deposits, and exception resolution still require human action, capping full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Financial institutions and many organizations require audit trails, internal controls, and approval workflows that create friction; regulatory oversight (SOX, banking regulations) and organizational risk policies demand human sign-off on certain disbursements, though routine collections and reconciliation face fewer barriers. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement for this role, but internal controls, segregation-of-duties requirements, and accountability for cash handling create moderate organizational friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven accounting automation (cloud RPA, integrated ERP modules) costs a fraction of loaded secretary wages ($45k–$60k annually); a mid-tier system can handle hundreds of transactions monthly for a few thousand dollars per year, achieving an order-of-magnitude cost advantage. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Software subscriptions are cheap per transaction, but integration, exception handling, and required human oversight for accuracy and fraud prevention keep total cost roughly comparable to a fraction of a secretary's time rather than an order of magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature accounting software and RPA platforms (UiPath, Blue Prism, SAP automation) reliably perform money collection, fund disbursement, and basic reconciliation in production systems across enterprises. Error rates on routine transactions are low, though complex or ambiguous items may still require human intervention. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Accounting automation tools (e.g., QuickBooks, Bill.com) reliably handle disbursements and reconciliation in production, but physical money collection/deposit and error correction still commonly involve human oversight. |
Establish work procedures or schedules and keep track of the daily work of clerical staff.
57CI 35–79 · exposure 50 · augmentation 75 · importance 3.8/5 · click for rater detail
Establish work procedures or schedules and keep track of the daily work of clerical staff.
57| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Fast adoption is underway in information and professional-services sectors; mid-market companies routinely deploy workflow automation and AI-driven scheduling tools. Manufacturing and smaller firms lag, but the trend is firmly toward automation in digitized workplaces. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Administrative/clerical management functions in typical offices adopt AI tools slowly, mostly through incremental scheduling or tracking software rather than full workflow management agents. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI scheduling and work-tracking tools significantly boost administrative assistant productivity by automating routine monitoring and flagging anomalies, freeing them to focus on strategic planning and staff development while they remain in supervisory control. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling assistants, task trackers, and dashboards can meaningfully improve efficiency in monitoring workflows and setting schedules, giving strong augmentative value even though full automation is unlikely. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Most of this task—scheduling, tracking work completion, monitoring staff activity, generating reports—can be automated via AI agents that read calendars, task management systems, and work logs. However, interpersonal aspects (coaching underperformers, adjusting for complex team dynamics) require some human judgment, preventing a full 5. |
| Task automatability | claude-sonnet-5 | 2/5 | Setting work procedures and tracking clerical staff daily activity requires ongoing judgment, interpersonal coordination, and adaptation to office dynamics that AI cannot fully replace, though scheduling and tracking sub-components could be assisted.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist; no licensing required. However, organizational culture, manager preference for human oversight, and staff resistance to automated monitoring create moderate friction against full displacement. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No strict licensing requirement, but organizational trust, accountability for staff performance, and preference for human oversight of personnel create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automating scheduling and work tracking via cloud-based agents costs a small fraction of a full-time administrative salary once integrated into existing systems, and scales with near-zero marginal cost per additional worker monitored. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While scheduling tools are cheap, the managerial oversight, exception handling, and interpersonal supervision components still require human involvement, keeping overall cost comparable to or only modestly cheaper than a human. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Production systems like Zapier, Make, and enterprise workflow platforms already automate schedule creation and work tracking at scale. Calendar and task-management integrations are mature. Minor gaps remain in real-time adaptive scheduling and nuanced delegation, but deployed products reliably handle core functions. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some workforce management and scheduling software exists but reliable end-to-end management of staff work tracking and procedure-setting by AI in production is limited and typically supervised by a human coordinator. |
Greet visitors or callers and handle their inquiries or direct them to the appropriate persons according to their needs.
56CI 50–61 · exposure 42 · augmentation 75 · importance 4.3/5 · click for rater detail
Greet visitors or callers and handle their inquiries or direct them to the appropriate persons according to their needs.
56| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Many organizations across information, finance, and professional services sectors have already deployed automated call screening, IVR systems, and chatbots for initial visitor/caller handling. Adoption is measurable and growing, though complete replacement remains uncommon. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | AI receptionist and call-routing tools are increasingly adopted in office and service settings, though many organizations still rely on human staff, especially for physical front-desk presence. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered CRM systems and caller intelligence tools routinely assist human receptionists by pre-filling context, suggesting routing, and alerting staff to incoming priorities. This augmentation significantly raises human productivity in handling mixed inbound flows. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI chatbots, call routing, and scheduling assistants significantly reduce routine inquiry burden for administrative assistants, letting them focus on more complex visitor needs. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can handle basic caller routing and scripted inquiries via chatbots, complex visitor needs assessment and judgment-based routing to the right person typically requires human contextual understanding and relationship awareness. Current systems cannot reliably handle the full interpersonal and organizational knowledge required end-to-end. |
| Task automatability | claude-sonnet-5 | 3/5 | AI phone/chat agents can route calls and answer routine inquiries, but in-person greeting and nuanced routing decisions still require significant human judgment, so only partial time savings are achievable today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or licensing barriers exist for automating reception tasks; companies face mainly organizational and customer-experience friction (clients may prefer human greeting). No legal requirement mandates human staffing for initial greeting or routing. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for this task, but organizational and customer preference for a human presence, especially for in-person visitors, creates moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven call routing and initial triage is substantially cheaper than human receptionist labor once deployed, though integration and oversight costs apply. The loaded wage for a full-time receptionist far exceeds the per-interaction cost of automated systems. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI phone/chat receptionist services cost a fraction of a human's loaded wage for routine inquiry handling and routing, though some oversight and escalation paths remain necessary. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Deployed phone systems and chatbots perform basic call screening and routing in production, but they have material error rates on ambiguous inquiries and lack the ability to assess visitor intent with nuance. Success is narrow and often requires human handoff. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Virtual receptionists, chatbots, and IVR/AI phone systems are deployed in production for call handling, but they still have material error rates and struggle with in-person greeting and edge-case routing. |
Coordinate conferences, meetings, or special events, such as luncheons or graduation ceremonies.
33CI 30–35 · exposure 25 · augmentation 75 · importance 3.5/5 · click for rater detail
Coordinate conferences, meetings, or special events, such as luncheons or graduation ceremonies.
33| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Event coordination remains largely handled by humans in organizations of all sizes; adoption of AI agents for this task is minimal. While larger corporations may trial scheduling automation, production-level autonomous event coordination is rare across the sector. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Administrative support roles are adopting scheduling and calendar AI tools, but event coordination specifically remains a slower-adopting, high-touch task within this occupation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist coordinators by automating reminders, tracking RSVPs, generating agendas, managing vendor communication templates, and flagging scheduling conflicts. These augmentations can materially reduce planning friction while the human retains decision authority and client relationships. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling assistants, calendar tools, and communication drafting significantly speed up logistics coordination even though a human remains essential for decision-making and on-site execution. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with scheduling, participant communication, and logistics planning, coordinating conferences requires real-time problem-solving, stakeholder management, and contingency handling that current systems cannot fully automate. The task involves negotiation, conflict resolution, and adaptive decision-making that fall short of the 50% time-saving threshold for end-to-end automation. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can assist with scheduling, invitations, and logistics tracking, but coordinating conferences/events involves vendor negotiation, in-person problem-solving, and real-time adjustments that current AI cannot fully handle end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Events often involve client or stakeholder preferences for human contact and decision-making, and liability concerns if an automated system makes poor vendor or schedule choices. However, no licensing requirement or hard legal barrier prevents AI from handling parts of the coordination workflow. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but organizational preference for a human point of contact and the need for judgment in vendor/venue negotiations create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Coordination requires human judgment, communication, and oversight that current AI integration does not significantly reduce in cost. Integration, prompt engineering, and human verification of decisions needed for event success make AI comparable to or more expensive than hiring an assistant for straightforward coordination work. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can reduce time on scheduling and communications, but human coordination, vendor relationships, and on-site management still require paid staff time, keeping costs comparable rather than dramatically lower. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably coordinates entire conferences or special events autonomously. AI can handle narrow subtasks like sending invitations or scheduling, but orchestrating multiple vendors, managing attendee needs, and handling day-of logistics remain largely manual; products exist only for fragments of this work. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Scheduling assistants and event-planning tools exist but are narrow point solutions; no deployed product manages full event coordination reliably without significant human oversight. |
Manage projects or contribute to committee or team work.
32CI 32–32 · exposure 25 · augmentation 75 · importance 3.6/5 · click for rater detail
Manage projects or contribute to committee or team work.
32| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Mid-market and enterprise organizations have adopted AI-assisted project tools and meeting automation, but true autonomous project management remains rare. Adoption is substantial in digital-forward sectors (tech, finance) but much slower in less digitized settings, placing overall velocity in the middling range. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support roles are adopting AI productivity tools steadily but project/team management remains largely human-led with AI as a support tool. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at augmenting project work: automating scheduling, summarizing meetings, flagging risks, surfacing data, and generating status reports all materially lift human productivity. The human remains central for judgment, but AI assistance on routine coordination substantially reduces administrative burden. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly help with meeting summaries, task tracking, scheduling, and drafting communications that support project and committee work, boosting the human's efficiency. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with administrative coordination (scheduling, document management, basic tracking), true project management requires dynamic decision-making, stakeholder negotiation, priority arbitration, and adaptive planning that current systems cannot do end-to-end with 50%+ time savings. Most of the value lies in judgment and human coordination, not routine execution. |
| Task automatability | claude-sonnet-5 | 2/5 | Project management and committee contribution require coordinating people, judgment calls, and interpersonal facilitation that current AI cannot fully replace, though it can assist with scheduling, notes, and tracking. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Organizational friction and accountability concerns are moderate: teams often prefer human coordinators for stakeholder communication and conflict resolution. There is no legal licensing requirement, but liability for project failures and the expectation of human judgment create some friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but organizational trust, accountability for decisions, and the need for human representation in team/committee settings create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI project management aids (scheduling bots, meeting summarizers) cost modest sums but require significant human oversight and decision-making. The fully-loaded cost of AI infrastructure plus the human supervision needed remains comparable to or exceeds simple human administrative work. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can reduce time spent on administrative overhead within project work, but a human still must be involved in decision-making and coordination, limiting overall cost savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Project collaboration tools (Asana, Monday.com) with AI-powered suggestions exist in production, but they augment rather than replace human managers. Current AI cannot autonomously manage scope changes, resolve team conflicts, or make trade-offs that humans must ultimately approve. Deployments are narrow and oversight-heavy. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some project management software includes AI features for tracking tasks and summarizing status, but no deployed product autonomously manages projects or represents a person on a committee. |
Supervise other clerical staff and provide training and orientation to new staff.
29CI 25–32 · exposure 25 · augmentation 63 · importance 3.7/5 · click for rater detail
Supervise other clerical staff and provide training and orientation to new staff.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Even digitized sectors retain human supervisors; there is no measurable displacement of supervisory roles by AI agents in production, and organizational reluctance to automate management functions keeps adoption minimal. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Administrative and office support roles are seeing moderate AI tool adoption for drafting and scheduling, but supervisory functions specifically lag behind due to their interpersonal nature. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist supervisors by automating scheduling, generating training agendas, tracking compliance checklists, and flagging performance metrics, materially improving administrative efficiency while the human supervisor retains decision-making. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully help by drafting onboarding materials, training schedules, checklists, and answering routine staff questions, freeing the supervisor to focus on interpersonal aspects. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can help draft training materials and schedule onboarding, the core function of supervision—providing feedback, mentoring, conflict resolution, and real-time guidance—requires human judgment and interpersonal presence that current systems cannot reliably replicate end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | Supervision and training of staff require interpersonal judgment, motivation, real-time feedback, and relationship management that current AI cannot fully replicate end-to-end.dmin AI can assist with materials but not perform the actual supervisory relationship. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Supervision carries inherent accountability for staff performance, legal compliance, and culture; organizations and employees expect human supervisors, and liability concerns around delegating people management to AI systems create strong adoption friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but organizational norms, accountability structures, and the need for human judgment in performance management create real friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools (draft templates, scheduling assistants) reduce administrative overhead but cannot replace the supervisor's wage; all-in cost remains well above AI alternatives because human oversight is still required. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can cheaply generate training documents, but the supervisory component still requires a human manager's time, so overall cost savings versus a human supervisor are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production systems today reliably supervise human staff or conduct training interactions independently; AI can support documentation and messaging but cannot own the supervisory relationship or accountability. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some AI tools exist for onboarding content generation or training modules, but no deployed product independently supervises staff or manages performance in production. |
Operate office equipment, such as fax machines, copiers, or phone systems and arrange for repairs when equipment malfunctions.
28CI 21–35 · exposure 20 · augmentation 38 · importance 4.1/5 · click for rater detail
Operate office equipment, such as fax machines, copiers, or phone systems and arrange for repairs when equipment malfunctions.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Most organizations still rely on humans for equipment troubleshooting and repair coordination; even in digitized sectors, these tasks remain on the periphery of automation efforts and adoption has been slow and limited. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Administrative support roles are seeing AI adoption in scheduling and communications, but equipment operation and repair coordination are not areas of active AI deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by monitoring equipment status, logging issues, and suggesting nearby repair vendors, reducing the secretary's time spent on troubleshooting and coordination, though the human must still make decisions and communicate with vendors. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could help find repair vendors, draft repair request emails, or troubleshoot via documentation lookup, but this offers only marginal assistance to the core physical task. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Operating standard office equipment (fax, copiers, phones) is straightforward and could be partially automated through robotic process automation or device APIs, but identifying and arranging repairs requires human judgment and communication—a full end-to-end automation meeting 50% time savings at equal quality is unlikely because the diagnostic and vendor-coordination components still demand human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | The physical operation of office equipment and coordinating repairs requires physical presence, dexterity, and phone/vendor coordination that current AI cannot perform end-to-end without a human physically present. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Equipment operation and repair coordination are not legally restricted or heavily regulated, but many organizations prefer human judgment in vendor selection and on-site troubleshooting; customer preference and the need for in-person diagnostics create moderate friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal barrier exists, but the inherent physical nature of operating machines and dealing with technicians on-site creates a structural barrier to remote AI substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The cost of AI-driven monitoring and RPA systems for equipment management, combined with required human oversight of repairs and vendor negotiation, approaches or exceeds the cost of a secretary handling these tasks part-time, making the ratio unfavorable for automation. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | AI cannot substitute for the physical operation and repair coordination, so there is no viable AI cost comparison for the core physical task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While some aspects—scheduling, logging issues, sending repair requests—can be handled by task automation software, no deployed product reliably diagnoses equipment faults and independently arranges repairs without human verification; most real-world solutions still require a person to troubleshoot and coordinate. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product operates physical office machines or manages on-site repair calls; this remains a physical-world task outside current AI product scope. |
Learn to operate new office technologies as they are developed and implemented.
25CI 13–38 · exposure 13 · augmentation 75 · importance 3.5/5 · click for rater detail
Learn to operate new office technologies as they are developed and implemented.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Organizations increasingly use AI-powered learning platforms and documentation assistants, but adoption remains uneven and often focused on initial onboarding rather than continuous learning of emerging technologies. Widespread production-scale displacement is not yet evident. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Administrative support roles show moderate AI tool adoption for specific subtasks like scheduling or drafting, but this particular meta-task of learning new tools isn't itself an adoption target. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI substantially assists this task by generating personalized training summaries, answering questions about new feature sets in real-time, creating practice scenarios, and reducing time spent searching documentation—all while the human remains the decision-maker and learner. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tutoring, chatbots, and interactive help systems can significantly speed up how quickly a secretary learns new software or office technology, functioning as an on-demand training aid. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Learning to operate new office technologies requires adaptive learning and contextual understanding that varies significantly per technology. While AI can assist with documentation parsing and training content generation, the task fundamentally demands human judgment about which features apply to individual workflows and hands-on experimentation—neither of which AI can fully automate or replace at the required depth. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a continuous learning/adaptation task performed by a human, not a discrete output that AI can generate or replace; AI cannot 'learn' new tools on behalf of the employee in a way that substitutes for their own skill acquisition.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While organizational preferences and the need for human validation create some friction, there are no strict licensing or legal barriers that prevent AI from assisting with or partially automating the delivery of training materials and initial familiarization steps. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or regulatory barrier exists, but the task is definitionally tied to human cognitive adaptation and organizational onboarding processes, giving it mild structural protection from automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-generated training content or documentation summaries may reduce the time to find information, but the core learning activity still requires human engagement. The cost of AI assistance here is not negligible relative to the human time spent, since the human must still do the actual learning and practice. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Since AI cannot perform this task at all, there is no meaningful AI cost to compare against human wage for this specific activity. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably and independently learns, then demonstrates mastery of new office technologies across diverse organizational contexts. AI chatbots can retrieve documentation and explain features, but they cannot perform the exploratory, adaptive learning that constitutes the actual task, nor validate competency in real-world usage. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | There is no deployed product whose function is to perform an employee's own on-the-job learning of new technologies; this is inherently a human capability-building activity. |
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.