All roles
for founders

AI for founders

As a founder you're doing the work of 5 specialists. AI doesn't replace any of them, but it lets you go from "I can't" to "I can sketch a first draft" across every function. The lever isn't doing things faster. It's being able to do things at all.

8 use cases·7 tools·30-min starter
Get the Founder skill pack (11)Hand-written workflows you install in ~/.claude/skills/

What AI handles well

Customer interview synthesis

The problem. You ran 8 customer interviews this week. Each is a 45-minute transcript. The patterns are in there but extracting them takes hours of re-listening.

What AI does. Paste transcripts into AI. Ask for: the recurring pain words, the workarounds people use today, the moments of strong emotion. The patterns surface in 10 minutes instead of an afternoon.

First-draft strategy docs

The problem. You have rough notes about a strategy decision. Turning them into something readable takes a half-day you don't have.

What AI does. Feed messy notes into the strategy doc prompt. Get a structured first draft. Edit the parts only you can write. Ship in an hour instead of a half-day.

Investor updates that actually get read

The problem. Monthly investor update is overdue. You hate writing them. Investors skim them anyway.

What AI does. Use a structured update template (wins, asks, lowlights, key metrics). Have AI draft from your raw bullet points. Spend the saved time on the "ask" section, which is the only part investors actually act on.

Founder-led sales when you're the only seller

The problem. You're running early-stage outreach yourself. You're not naturally a salesperson. Generic templates feel cringe.

What AI does. Use the cold outreach prompt with your founder voice. Write 3 emails per day, send them yourself, watch what works. AI helps you not sound like you're reading a script.

Hiring: JDs, screens, evaluations

The problem. You need to hire your first marketing person but you've never written a marketing JD. You don't know what to ask in screens. Calibrating "good" is hard.

What AI does. Use the JD prompt to write something candidates will actually read. Use AI to draft screening questions specific to the role. Have it help you debrief candidates by structuring your notes against the rubric.

Decision frameworks for hard calls

The problem. You're deciding whether to raise, hire, kill a feature, or fire a customer. You're going in circles.

What AI does. Use the decision framework prompt to structure the choice — reversibility, worst case, friend test, smallest-version-that-earns-info. Don't outsource the decision; use AI to make sure you're asking the right questions.

Rapid prototyping with no-code or vibe-code

The problem. You want to test a hypothesis. Building a real prototype takes 2 weeks of eng time you don't have. Lovable / v0 / Cursor agents can ship a clickable demo in a day.

What AI does. Skip the figma → spec → eng pipeline for hypothesis testing. Use AI builders to ship a clickable mock or even a working demo by Friday.

Tools:Lovable for full-stack apps. v0 for landing pages and components. Cursor for hand-coding with AI assistance.

Bug triage and customer support overflow

The problem. A customer reports a bug on Sunday. You're the only one who can respond. You don't have time to write a thoughtful reply, but the customer matters.

What AI does. Use AI to draft a customer-facing response that acknowledges the issue, sets expectations, and asks for the info you'll need to debug. You edit and ship in 2 minutes.

Your AI stack

Start with the foundation. Add specialized tools as the work calls for them.

Foundation LLM

Claude
Best for long-form writing, nuanced thinking, and code. The default LLM for founders doing serious work, not just task automation.
ChatGPT
Wider tool integration. Strong for structured tasks, image generation, voice. Worth having alongside Claude.

Specialized add-ons

Cursor
AI-assisted coding even if you're not a full-time engineer. Lets you ship fixes and small features without bothering the team.
Lovable / v0
Build clickable prototypes without writing code. Test hypotheses in days instead of weeks.
Notion AI
Drafts and summaries in your existing docs. Where most founder writing lives anyway.
Granola or Otter
AI meeting notes. Frees you from typing during calls so you can actually be present.
Perplexity
Research + citations. For when you need actual fresh info, not LLM training data.

Prompts ready to use

Get started in 30 minutes

1

Set up one Claude Project per "hat" you wear

15 min

One for founder-CEO writing (investor updates, strategy docs). One for founder-marketer (positioning, content). One for founder-engineer if technical. Each loaded with relevant context.

2

Use the weekly review prompt this Friday

10 min

Run the conversational weekly review. It surfaces what you're actually working on vs what you should be. Most founders are surprised how much low-leverage work they're doing.

3

Pick one repetitive task you've been avoiding and have AI draft it

5 min

Investor update? Customer interview synthesis? Job description? The one you've been pushing for two weeks. Get a first draft today.

Common mistakes

  • Outsourcing thinking on hard decisions. AI is a thinking PARTNER, not a decision-maker. Use it to ask better questions, not to give you answers you'll act on without reflection.

  • AI-generated investor updates that read like every other AI-generated update. Investors notice. Your authentic voice signals what AI's default doesn't: that you're actually paying attention.

  • Hiring a "Head of AI" too early. What you usually need is each function learning to use AI well, not a centralized owner. Centralization comes later.

  • Believing AI can replace user research. AI summarizes transcripts brilliantly. It cannot tell you what to ask in the next interview. That's judgment, not analysis.

  • Using AI to write something you'd normally not write at all. If you wouldn't write a 5-page strategy doc by hand, generating one with AI just creates more docs no one reads.

Related roles