AI for Executive Assistants That Knows the Exec

Updated January 2026 | 6 min read

You're managing the CEO's calendar. A board member wants a meeting. You ask AI to draft a scheduling email.

It gives you generic corporate politeness: "Would next Tuesday at 2pm work for you?"

Wrong on three levels. The CEO doesn't take meetings after 3pm. Board members get executive-level formality, not casual tone. And this particular board member prefers morning calls, not calendar invites.

You know all this because you've been the EA for two years. AI has no idea.

You re-write the email. Tomorrow, different stakeholder, same problem — AI doesn't know the preferences, the protocols, or the relationships. You're not saving time. You're training AI from scratch every single day.

Executive assistants don't need generic scheduling help. They need AI that knows the exec as well as they do.

Why Generic AI Fails Executive Assistants

ChatGPT doesn't know your executive. It doesn't know their calendar rules, communication style, travel preferences, or the unwritten protocols that keep their day running.

Executive assistants manage:

• Calendar preferences (no meetings before 9am, blocks of deep work on Tuesdays, 15-minute buffer between calls)
• Meeting protocols (board members get 60 minutes, vendors get 30, internal teams get 15 unless flagged urgent)
• Communication style (email tone shifts by stakeholder, some people get formal, some get casual)
• Travel standards (aisle seat, Marriott properties only, no red-eyes, car service not Uber)
• Stakeholder relationships (who gets priority, who gets rescheduled, who the exec actually likes talking to)
• Decision-making quirks (won't commit to evening events more than two weeks out, always says yes to customer calls, hates video meetings before coffee)

When you ask AI to draft an email, schedule a meeting, or book travel, it has none of this. It can't write in the exec's voice because it doesn't know the voice. It can't prioritize calendar requests because it doesn't know the relationships.

The AI isn't bad at assisting. It's uninformed.

What Executive Assistants Actually Need

You need AI that remembers:

The exec's operating manual. When they take meetings, how long meetings should be, what days are protected for deep work, what time zones they're in this week.

Communication style by stakeholder. Board members get formal. Direct reports get warm but direct. Customers get enthusiastic. Vendors get polite but brief. Tone shifts matter.

Travel preferences. Not just "aisle seat" — the full stack. Preferred airlines, hotel chains, car service companies, dietary restrictions, meeting room setup preferences, AV needs.

Stakeholder relationships. Who gets priority access to the calendar. Who can be rescheduled. Who the exec avoids unless absolutely necessary. Who gets a personal call back vs. a delegated response.

Meeting prep requirements. What the exec needs before a board meeting (full deck, financials, talking points). What they need before a customer call (account history, recent interactions, pain points). What they don't need (long briefing docs they won't read).

Generic AI can't do this. It needs context files.

How Context Files Work for Executive Assistants

Context files are markdown documents that live in Obsidian. AI reads them every time you start a conversation.

One file might be exec-preferences.md:

• Calendar rules (meeting hours, buffer times, protected blocks)
• Communication preferences (email vs. call vs. text, response time expectations)
• Decision-making patterns (what they approve immediately, what needs more info)
• Personal quirks (doesn't read long emails, prefers bullet points, hates small talk)

Another might be meeting-protocols.md:

• Meeting length by stakeholder type
• What gets scheduled vs. what gets delegated
• Prep materials needed per meeting type
• Follow-up expectations

Another: stakeholder-map.md:

• Key people (board members, direct reports, top customers, key vendors)
• Relationship status (priority access, normal queue, low priority)
• Communication style per person
• Context notes (current projects, recent interactions, preferences)

When you ask AI to draft a scheduling email for a board member, it reads exec-preferences.md (knows calendar rules), meeting-protocols.md (knows board meetings are 60 minutes), and stakeholder-map.md (knows this board member prefers morning calls and gets formal tone).

First draft is right. You don't re-explain preferences. You don't adjust tone. You just send.

Before and After

Before: "Draft a scheduling email for the board meeting with Jennifer."
AI gives generic corporate: "Would Tuesday at 2pm work?"

You re-write it. Jennifer is a board member, so formal tone. She prefers morning calls. The CEO doesn't take meetings after 3pm anyway. Board meetings need 60 minutes, not 30. You know all this. AI doesn't.

Every email, you're correcting the same mistakes.

After: "Draft a scheduling email for the board meeting with Jennifer."
AI reads stakeholder-map.md (Jennifer is a board member, prefers mornings, gets formal tone), meeting-protocols.md (board meetings are 60 minutes), and exec-preferences.md (CEO available 9am-3pm). First draft: "Good morning Jennifer, would you be available for a 60-minute board call on Tuesday at 10am?" Done.

Next request: "Book the CEO's travel to the Austin conference."
AI reads travel-preferences.md, books the aisle seat on the preferred airline, Marriott property downtown, car service from the airport, books the meeting room with AV setup. You review and confirm. Took two minutes.

No re-explaining. No correcting. AI remembers.

What This Looks Like in Practice

An executive assistant to a CEO sets up five context files:

1. exec-preferences.md — Calendar rules, communication style, decision patterns, quirks
2. meeting-protocols.md — Meeting lengths by type, prep requirements, follow-up rules
3. stakeholder-map.md — Key people, relationship priority, communication preferences
4. travel-standards.md — Flight preferences, hotel chains, car service, dietary needs
5. current-priorities.md — What the exec is focused on this month, what gets priority, what gets deferred

Total setup time: one afternoon.

Now when she asks AI to draft emails, schedule meetings, or book travel, it knows the preferences. When priorities shift (new product launch takes precedence, board meeting prep becomes urgent), she updates current-priorities.md. AI adjusts automatically.

The context files become the exec's operating manual. New EAs use them for onboarding. AI uses them for every task. The exec never has to explain preferences twice.

What You Get

AI that drafts emails in the right tone for the right stakeholder without being told.

AI that schedules meetings according to the exec's actual calendar rules and stakeholder priorities.

AI that books travel matching every preference (flights, hotels, car service, meal needs).

AI that preps meetings with the right materials based on who's attending and what's being discussed.

AI that gets smarter as you update preferences, relationships, and priorities.

No more re-explaining the exec's preferences. No more correcting tone and timing. No more starting from zero every task.

Executive assistants already maintain the exec's operating manual in their head. This just makes AI read it.