The best ChatGPT prompts for founders

By the Chat Judge team · Updated 2026-10-19 · 5 min read · Links go to prompts in our library; creators are credited on each prompt.

In short: The prompts worth keeping are the ones you run every week: a follow-up after a meeting, a proposal deck from your own references, landing page copy from facts you supply, and a cold email in a few lengths. This guide walks through those, with a link to each prompt in the library so you can copy it rather than retype it.

Start with the emails you send every week

Most of a founder's ChatGPT use is email, and the quality gap is usually not the model. It is that the prompt does not say who the reader is, what was agreed and what the next step should be. Meeting follow-up email fixes that by making you paste your notes, your recommendation and one agreed next step, and it tells ChatGPT to use only the facts you gave it. That last line matters more than the wording: it is what stops invented dates and promises.

Refine an email is the second one worth pinning. You paste a rough draft and ask for it to be clear, natural and professional without losing the details. It is deliberately small, which is why it works: you are not asking ChatGPT to decide your message, only to clean it up.

Two more cover the rest of the week. Email that sends a proposal writes the short note that goes with an attachment and invites a conversation, and Introduce your company by email turns a generic intro into one that highlights the use cases relevant to that specific company. Both expect you to bring the facts.

Decks: give ChatGPT the references, not a blank page

A deck prompt fails when it starts from nothing, because ChatGPT will fill the gaps with claims you cannot defend. Create a customer proposal deck starts the other way: you name the customer, the requirements, the audience and you attach your own reference material, and it asks for a story built from those. It includes style instructions for tools that actually build slides, so you get an outline you can hand to one.

Once a deck exists, the useful prompts are the editing ones. Make a deck suitable for senior management reshapes a working deck for a CEO, business leaders and a CTO, moving technical detail into an appendix. Pilot results deck for a customer CEO does the same for a finished pilot, asking for results, operational gaps and a recommendation for the next phase, using the analytics you attach. And when a long document needs to survive a meeting, Turn a document into a one-slide summary forces one takeaway instead of a compressed wall of text.

Landing pages and the page that is not converting

Write a landing page asks for the pieces a page needs: headline, supporting copy, benefits, how it works, proof, FAQs and one call to action, built from facts you paste in. The proof section is where people get into trouble. If you have no testimonials or numbers, say so in the prompt and tell ChatGPT to mark the spot rather than invent one.

When a page already gets traffic, Improve landing-page conversion is the better starting point. It reviews the offer, clarity, trust and call to action, ranks the problems and rewrites the weak sections. Treat the list as hypotheses. ChatGPT cannot see your analytics, so only testing with visitors tells you which change matters. If the problem is that your ad says one thing and the page says another, Align an ad with its landing page checks that the audience, promise, offer and call to action match.

Proposals, outreach and a note on what is shared

Pilot proposal from your notes turns meeting notes into a proposal with objectives, scope, channels, success metrics, pricing and next steps. Pair it with the email prompt above when you send it.

For first contact, Cold email in three lengths was shared on X by AImonkey. It asks for three versions of one email, at 50, 80 and 100 words, with a subject under five words and a single question as the call to action. Having three lengths is useful because you can see which one still makes sense when you cut it, and you should still check that the proof it uses is proof you actually have.

Finally, Facebook campaign and ad copy plans a month of paid social: frequency, messages and copy from the packages and prices you provide. It is a planning prompt, not a performance promise, and the copy it produces is a draft to test.

A small system for reusing them

Ten prompts you can never find again are worth less than three you use. Keep a short document, one prompt per heading, and paste the founder context block once at the top of each new chat so you stop re-explaining your company. When a result is genuinely good, do not just copy it: ask ChatGPT what in the prompt made it work, and save that version.

Two smaller prompts round out a week. Improve a single slide fixes the one crowded slide instead of regenerating the deck, and Check a proposal against customer requirements compares the finished deck with what the customer actually asked for before you send it. Run that last one every time. It catches the slide you assumed was implied.

For announcements and handoffs, Partnership announcement on LinkedIn and Turn a campaign plan into a marketing brief are the same idea applied to other audiences: you provide the facts and the relationships, the prompt provides the structure, and a person checks the names.

How to get good results from any of these prompts

1. Fill in every placeholder. Anything in [BRACKETS] is a fact only you know. Leave one empty and the model will guess.
2. Tell it what not to invent. Add "use only the facts above, and flag anything missing" if the prompt does not already say it.
3. Ask for one version first. Fix the direction, then ask for variations.
4. Check every name, number and promise before anything leaves your hands.
5. Edit it into your own voice. Read it aloud once; if it does not sound like you, change it.

Prompts in this guide

EmailsChatGPTMeeting follow-up emailRight after a customer or partner meeting, when you have your notes and a clear next step.View prompt →EmailsChatGPTRefine an emailYou have a rough draft and want it clear, natural and professional.View prompt →EmailsChatGPTEmail that sends a proposalYou are attaching a final proposal and want it to start a conversation.View prompt →EmailsChatGPTIntroduce your company by emailA warm intro to a company or person who may need what you do.View prompt →SlidesChatGPTCreate a customer proposal deckYou are pitching an AI platform to a specific customer and have reference material to hand.View prompt →SlidesChatGPTMake a deck suitable for senior managementA working deck needs to be reshaped for the CEO, business leaders and CTO.View prompt →SlidesChatGPTPilot results deck for a customer CEOA pilot has finished and you must show results and win the next phase.View prompt →WebsitesChatGPTWrite a landing pageLaunching an offer and starting from a blank page.View prompt →WebsitesChatGPTImprove landing-page conversionA page gets traffic but not enough sign-ups or enquiries.View prompt →WebsitesChatGPTAlign an ad with its landing pageThe ad promises one thing and the page says another.View prompt →ProposalsChatGPTPilot proposal from your notesYou have meeting notes and need a customer-ready pilot proposal.View prompt →

EmailsChatGPTClaudeCold email in three lengthsAct as a cold email expert. My service: [service] Prospect: [name, role, company] Problem I…View prompt →AImonkey @AImonkey108Report or remove

MarketingChatGPTFacebook campaign and ad copyPlanning a month of paid social for an offer.View prompt →SlidesChatGPTTurn a document into a one-slide summaryYou need the whole point of a long document, dataset or notes on one slide.View prompt →

Keep exploring

More guides

• The best Claude prompts for founders
• The best Gemini prompts for founders
• The best Grok Bot prompts
• The best Perplexity prompts for market research
• Muse prompts for your calendar and inbox
• ChatGPT vs Claude for founders’ writing
• Prompts for sales follow-ups
• Prompts for fundraising and investor updates
• Prompts for hiring your first employees

Written by the Chat Judge team from prompts in our library. We do not run model benchmarks, and nothing here is a claim about results you will get. Spot a problem? Tell us or submit a better prompt.