Secretaries and administrative assistants

Perform basic bookkeeping tasks

Basic bookkeeping—data entry, invoice processing, expense categorization, reconciliation—is highly automatable with current AI and RPA tools. OCR plus AI can extract data from receipts and invoices, categorize transactions, and flag anomalies with minimal human oversight needed.

BLS evidence: Listed as a typical duty but presented as one of several routine clerical tasks.

Maintain databases and filing systems

Database maintenance and digital filing are highly structured tasks that AI excels at—automated tagging, categorization, deduplication, and retrieval are mature capabilities. Physical filing requires robotics, but most modern offices are digital-first where AI can operate autonomously.

BLS evidence: Listed as a typical duty and organizational skills to 'keep files, folders, and schedules in order' are emphasized.

Schedule appointments and update event calendars

AI scheduling assistants can parse email requests, check calendar availability, resolve conflicts, and send confirmations autonomously. Tools like x.ai and Microsoft's scheduling AI already handle this end-to-end for routine appointments, though complex multi-party scheduling with unstated preferences may need human touch.

BLS evidence: Explicitly listed as a typical duty and emphasized as helping organizations run efficiently.

Arrange staff meetings

AI can identify optimal meeting times across calendars, send invitations, book rooms, and handle logistics. Current tools handle this with minimal human input for routine meetings, though meetings requiring political sensitivity about attendee selection or timing may need human judgment.

BLS evidence: Listed as a typical duty that supports organizational operations.

Answer telephones and take messages or transfer calls

AI phone systems can now handle call routing, take structured messages, and transcribe voicemails with high accuracy. Most routine call handling is automatable, though complex judgment calls about urgency or escalation may need human review.

BLS evidence: Listed first among duties that secretaries and administrative assistants typically do.

Handle incoming and outgoing mail and faxes

AI can sort, categorize, and route incoming digital mail/faxes, extract key information, and flag items for attention. Outgoing mail can be templated and auto-generated. Physical mail handling requires human involvement, but the cognitive work of processing correspondence is largely automatable.

BLS evidence: Explicitly listed among the duties secretaries and administrative assistants typically do.

Prepare documents, reports, presentations, and correspondence using computer software

AI can generate first drafts of routine documents, reports, and correspondence from templates or brief instructions, and format presentations. However, documents requiring organizational context, tone calibration for specific relationships, or strategic messaging still benefit from human drafting or substantial editing.

BLS evidence: The page states they 'use computer software to create spreadsheets; manage databases; and prepare presentations, reports, and documents.'

Edit documents

AI editing tools can catch grammar, style inconsistencies, and suggest improvements at a level matching or exceeding median human editors for routine documents. However, editing requiring deep organizational context, brand voice consistency, or strategic messaging judgment still benefits from human review.

BLS evidence: Listed among typical duties and requires writing skills to 'ensure accuracy.'

Conduct research and review incoming documents

AI can search, summarize, and extract key points from documents efficiently. However, research requiring judgment about relevance to unstated organizational priorities, or synthesis across disparate sources for novel insights, still benefits significantly from human direction and validation.

BLS evidence: Executive secretaries 'often handle complex responsibilities, such as reviewing incoming documents, conducting research, and preparing reports.'

Negotiate with vendors, buy supplies, and manage stockrooms or corporate libraries

While AI can compare vendor quotes and manage inventory databases, negotiation requires relationship management and judgment about trade-offs that AI cannot fully replicate. Physical stockroom management requires presence. AI assists substantially but humans remain load-bearing for vendor relationships and physical organization.

BLS evidence: The page states 'They also may negotiate with vendors, buy supplies, and manage stockrooms or corporate libraries.'