How to Use AI to Write a Project Proposal

You’ve spotted a problem at work, or an opportunity, and you’ve got an idea to fix it. Now you need someone to say yes, which usually means a proposal. Many good ideas never happen because the proposal was vague, too long or didn’t answer the questions decision-makers care about: why this, why now, how much and what’s the risk?

AI is a great partner for proposals. It can help you structure your thinking, write a crisp problem statement, set out options, outline a timeline and budget, and anticipate the questions you’ll be asked. The more context you give it, the better. Your knowledge of the organisation, the people and the numbers is what makes the proposal convincing. This guide takes about two hours for a short proposal.

What you’ll need

• A clear understanding of the problem or opportunity
• Any data you have: costs, time spent, complaints, sales figures
• Your organisation’s proposal template, if there is one
• An idea of who decides and what they care about
• An AI chatbot approved for work use, and about two hours

Step-by-step: write a project proposal with AI

1. Start with the problem, not the solution. Ask the AI to help you write a clear, specific problem statement with evidence (prompt 1). If decision-makers don’t agree there’s a problem, they won’t fund a fix.
2. Know your audience. Tell the AI who’ll read it and what they care about: cost, risk, customers, staff or strategy. Good prompts give context, as Anthropic’s prompt engineering guidance explains.
3. Structure it. Ask for an outline (prompt 2). Most proposals need a summary, problem, solution, options considered, benefits, timeline, costs, risks and next steps.
4. Fill it with real detail. Write each section using your data. Ask the AI to help turn rough numbers into a simple cost and benefit table.
5. Show you’ve thought about options. Include “do nothing” and one or two alternatives. It makes your recommendation more credible.
6. Stress-test it. Ask the AI to play a sceptical decision-maker and list their toughest questions (prompt 3). Add answers to the proposal.
7. Write the summary last. Ask for a one-page or half-page executive summary (prompt 4). Many readers only read this.

Copy-paste prompts

1. Sharpen the problem statement

I work as a [ROLE] at a [TYPE OF ORGANISATION]. The problem is [DESCRIBE]. It affects [WHO] by [IMPACT]. Evidence I have: [DATA, EXAMPLES]. The people deciding are [AUDIENCE] and they care most about [PRIORITIES].
Write a clear problem statement of 3 to 4 sentences that uses my evidence and connects to what the decision-makers care about. Then tell me what extra evidence would make it stronger.

2. Outline my proposal

Here's my idea: [DESCRIBE SOLUTION]. Rough cost: [COST]. Rough timeline: [TIMELINE]. Alternatives I considered: [OPTIONS].
Create an outline for a [LENGTH] proposal with headings for summary, problem, proposed solution, options considered (including doing nothing), benefits, timeline, budget, risks and mitigations, and next steps. Under each heading, list what I should include and any questions I still need to answer.

3. Play the sceptic

Here's my draft proposal: [PASTE DRAFT].
Act as a sceptical [SENIOR ROLE, SUCH AS FINANCE DIRECTOR]. List the 10 toughest questions you'd ask about cost, risk, timing, resources and evidence. Then tell me which sections of the proposal are weakest and how to strengthen them.

4. Executive summary

Write a one-page executive summary of this proposal for busy senior readers: [PASTE PROPOSAL].
Include the problem, recommendation, cost, benefits, timeline, key risk and the decision I'm asking for, in plain British English. Use short paragraphs or bullets.

Recommended tools

• Claude: strong at structuring long documents and playing devil’s advocate.
• ChatGPT: good for outlines, tables and summaries.
• Microsoft Copilot: helpful in Word if your organisation uses Microsoft 365.
• Gemini: useful in Google Docs.
• Grok: quick for brainstorming risks and options.

Tips and mistakes to avoid

• Lead with the decision you need. Say clearly what you’re asking for.
• Use real numbers. Don’t let AI invent statistics. Mark estimates as estimates.
• Keep it short. A clear five-page proposal beats a 20-page one. Put detail in appendices.
• Be honest about risks. Showing you’ve thought about what could go wrong builds trust.
• Follow your AI policy. Use approved tools and keep confidential data out.
• Talk to stakeholders first. Proposals rarely succeed as a surprise. Share early drafts.
• Make it skimmable. Clear headings, tables and bullet points help busy readers.

FAQ

How long should a project proposal be?

As short as possible while answering the key questions. For internal proposals, two to five pages plus a one-page summary is common.

Can AI write the whole proposal?

It can draft, but it doesn’t know your organisation, people or numbers. Use it to structure and polish, and supply the substance yourself.

What if my proposal is rejected?

Ask for feedback, find out what would change the answer, and revise. Timing, budget cycles and evidence often matter more than the idea itself.

Should I include a budget even if it’s rough?

Yes. Decision-makers will ask anyway. Give a range, explain your assumptions and say how you’ll firm it up, rather than leaving it out.

I work as a [ROLE] at a [TYPE OF ORGANISATION]. The problem is [DESCRIBE]. It affects [WHO] by [IMPACT]. Evidence I have: [DATA, EXAMPLES]. The people deciding are [AUDIENCE] and they care most about [PRIORITIES].
Write a clear problem statement of 3 to 4 sentences that uses my evidence and connects to what the decision-makers care about. Then tell me what extra evidence would make it stronger.

Related guides

• How to use AI to prepare a presentation and slides
• How to use AI to take meeting notes and summaries
• How to use AI to write professional emails faster

Reference: Claude Docs: Prompt engineering overview