Skip to content
Back to Resources
ClaudeExecutive SupportBeginnerGuide

Claude CoWork for Executive Assistants

A practical guide to using Claude as your AI co-worker for inbox triage, meeting prep, and follow-up execution.


What is Claude CoWork?

Claude CoWork is the practice of using Claude as a context-aware executive support partner inside your daily workflow. It is not just "ask AI to write an email." It is setting Claude up so it understands how your executive works, what gets escalated, what can be delegated, and how to turn a flood of context into a clean next step.

Claude-native prompts. The prompts in this guide use Claude's native XML tag structure (<context>, <instructions>, <format>, <avoid>) for more precise, consistent output. These tags help Claude parse your intent with less ambiguity. They also work in ChatGPT, but they are optimized for Claude.

This is where executive assistants are getting the most value from AI right now: inbox triage, meeting prep, follow-up drafting, and daily priority planning. The weak version of an AI assistant is a fancy note taker. The useful version helps you reduce context-switching and create momentum without losing judgment.

Setting Up Claude for Executive Support Work

The best setup is a dedicated Claude Project for executive support.

Step 1: Create a Project. In Claude, create a Project called something like "Executive Support Desk."

Step 2: Add custom instructions. Use something like:

You are my executive support assistant. Here is my working context:

<executive-profile>
- Executive role: [CEO / COO / Founder / Chief of Staff support]
- Priorities this quarter: [top 3 priorities]
- Communication style: [brief, direct, warm, formal, etc.]
- Escalation rules: [what should always be surfaced]
- Typical stakeholders: [board, customers, hiring managers, leadership team]
</executive-profile>

<working-rules>
- Assume I need concise answers first, details second
- Distinguish between executive work and coordination work
- When drafting replies, match tone without sounding robotic
- Always separate confirmed decisions from assumptions
- Always call out deadlines, owners, and unresolved questions
- Never invent context that was not provided
</working-rules>

Step 3: Upload reference material. Add your executive's preferred email examples, recurring meeting templates, planning docs, and any internal style guides.

Step 4: Stay inside this Project. The value comes from keeping your context persistent.

Your Top 5 Workflows with Claude

1. Inbox Triage

The biggest daily win is deciding what deserves the executive's attention and what does not.

<task>Triage this inbox summary for a CEO.</task>

<context>
- Executive only wants board, customer, hiring, and major financial items surfaced
- Everything else should be delegated or drafted if safe
- Tone should be concise and practical
</context>

<input>
- Board member requesting pre-read materials before tomorrow's call
- Customer escalation from enterprise account
- Recruiter asking for interview availability
- Vendor contract follow-up
- Internal budget question
- Three newsletters
</input>

<instructions>
- Group into: needs executive attention, delegate, can wait
- Draft short replies where useful
- Flag anything risky or sensitive
- Keep the output skimmable
</instructions>

Before Claude: 30-45 minutes to sort, summarize, and draft.
After Claude: 5-10 minutes to review and send.

2. Meeting Briefs

Meeting prep is one of the highest-leverage uses of AI for executive assistants.

<task>Create a prep brief for tomorrow's Q2 planning meeting.</task>

<context>
- Participants: COO, Head of Sales, Finance Director
- Topic: pipeline is behind target; hiring plan may need to be reduced by 10%
- Board update is due Thursday
</context>

<instructions>
- Summarize why this meeting matters now
- List likely friction points
- Give 5 talking points and 5 questions the executive should ask
- End with a prep checklist
</instructions>

Before Claude: 20-30 minutes chasing context.
After Claude: 5 minutes to sanity-check and personalize.

3. Follow-Up Drafts

The useful output is not the transcript. It is the recap that creates motion.

<task>Draft a follow-up email and action tracker from these notes.</task>

<notes>
- Pause two vendor evaluations
- Move roadmap review to next Wednesday
- Finance to model three headcount scenarios by Friday
- Need customer impact note before leadership review
</notes>

<instructions>
- Separate confirmed decisions from open questions
- Create an internal action table with owners and deadlines
- Write a polished recap email from the executive
- Keep it concise and easy to scan
</instructions>

4. Daily Priority Planning

This is where Claude becomes a real CoWork partner instead of a writing toy.

<task>Build today's priorities for a founder.</task>

<context>
- Investor prep at 8:30
- Three hiring interviews
- Customer escalation at 2:00
- Dinner with partner prospects
- 37 unread emails
</context>

<instructions>
- Give the top 5 priorities
- Suggest what to delegate
- Flag anything unrealistic
- Show the order of operations
</instructions>

5. Stakeholder Drafting

Executive assistants often need to write in multiple tones in the same hour.

<task>Draft three versions of this message.</task>

<context>
Need to move a leadership sync because the executive was pulled into a customer issue.
Recipients:
- Internal leadership team
- External partner
- Candidate in interview loop
</context>

<instructions>
- Keep each version appropriate to the audience
- Preserve the same core facts
- Sound like a human, not a template
</instructions>

Prompt Engineering Tips for Executive Assistants

1. Tell Claude what gets escalated. Otherwise it will summarize everything with equal weight.

2. Distinguish decision support from drafting. "Write the email" and "tell me whether this needs the executive" are different jobs.

3. Ask for owners and deadlines explicitly. If you do not ask, most AI output stays fuzzy.

4. Use examples of the executive's tone. Even two or three good emails improve drafting quality significantly.

5. Ask Claude to separate certainty from assumption. This is especially important in scheduling and follow-up work.

6. Keep inputs de-identified where possible. Use roles, not private personal details, unless your policy explicitly permits the tool.

Privacy & Discretion

Executive support work often includes sensitive information. Be stricter than average.

  • Do not paste personal data, compensation details, legal matters, health details, or board-sensitive information unless your organization's AI policy explicitly allows it.
  • Review every draft before sending. Claude should speed up judgment, not replace it.
  • Treat travel, security, investor, and personnel topics as higher-risk categories.
  • If you are using Claude for external communication, confirm names, dates, and commitments before the message leaves your outbox.

Going Further

Get weekly AI prompts for Executive Support professionals

Join professionals already saving hours every week. Free. No spam.