A ChatGPT Project Proposal: Prompt and Example

ChatGPT can help structure and draft a project proposal when you give it the right inputs: the goal, the scope, the constraints and the evidence you have available. Without those inputs, it will produce a generic template filled with confident-sounding guesses. The workflow below avoids that by asking the assistant to check its assumptions before drafting.

Once the proposal is approved, the decisions it contains need a home. Tools like fenbs record approved scope, constraints and open questions as Decisions and rules so any assistant you use later reads them before acting.

Gather your inputs before you prompt

Before opening ChatGPT, collect four things. First, the goal in one sentence: if you cannot write it in one sentence, the goal is not clear enough for a proposal. Second, the scope: list the three to five deliverables someone will be able to point to when the project is complete. Third, any evidence you have that the plan is feasible, such as past projects, vendor quotes or existing research. Fourth, the constraints: fixed deadline, spending limit, team size, required tools.

A prompt built on these four inputs produces a first draft that needs refinement. A prompt without them produces a draft that reads confidently but is built entirely from generic assumptions that may not fit your situation. Starting with the inputs takes ten minutes; discovering the wrong assumptions in a stakeholder review takes much longer.

What goes into a good project proposal

A proposal needs to answer four questions clearly: What is the goal? What is in scope and what is not? What evidence supports the plan? And what is still unknown? A proposal that skips the last two questions creates work later, when someone finds the missing evidence or discovers the unanswered question mid-project.

ChatGPT is good at structuring answers to the first two questions and at prompting you to gather evidence for the third. It cannot generate evidence for you. Any statistic, cost estimate or timeline it provides should be treated as a placeholder until you verify it from a primary source.

The executive summary is the hardest section to write well because it must make sense to someone who will not read the rest of the document. Write it last, after the scope and evidence sections are solid. Ask ChatGPT to draft the summary from those completed sections rather than asking it to draft a summary before you have the substance it needs to summarize.

A prompt for the first draft

Give the assistant the context it needs before asking it to draft. A one-paragraph brief is enough to get a useful first response.

You are helping me draft a project proposal.

Project: [short description]
Goal: [what success looks like]
Audience: [who will read this proposal]
Scope: [what is included]
Known exclusions: [what is explicitly not included]
Key constraints: [budget, timeline, technology, or other fixed limits]
Evidence I have: [any data, research or precedent I can cite]

First, list any assumptions you are making that I have not stated. Then draft the proposal with these sections: executive summary, goal and scope, approach, evidence and risks, open questions, and next steps.

The instruction to list assumptions before drafting is important. It surfaces places where the assistant is filling a gap with a guess. Review that list and either confirm the assumptions or replace them with your actual data before accepting the draft.

Checking scope and exclusions

After the first draft, run a second prompt specifically on scope. Scope creep often starts in the proposal itself, where a loosely worded objective creates room for later disagreement.

Review the scope section of this proposal. List anything that the goal statement implies but that I have not explicitly included or excluded. For each item, tell me whether it reads as in scope, out of scope, or ambiguous.

Work through the list and make a decision on each item. A proposal with clear exclusions is much easier to manage than one that relies on the reader to understand what was meant.

Handling open questions

Every proposal has questions that cannot be answered before approval: which vendor to use, whether a third party will cooperate, whether the budget will be confirmed. These belong in the proposal explicitly, not hidden in an optimistic assumption.

Review this proposal for any point where I have stated something as decided that may not yet be agreed. List each one as an open question with: what the question is, who needs to answer it, and by when it needs to be resolved to keep the project on schedule.

Keep open questions open in the final document. Do not let the assistant close them with a confident assertion. A reader who approves a proposal with stated open questions is more informed than one who approves a document that hides uncertainty.

Say a proposal for a new software system depends on an API from a third-party vendor. If you have not confirmed that the vendor's API supports the integration you need, that is an open question. Write it as: "Vendor API compatibility: pending confirmation from the vendor by [date]. If incompatible, the integration approach will need revision and may affect the timeline." That sentence is more useful than a confident claim that the integration will work.

Revising and strengthening the draft

Once you have addressed assumptions, scope and open questions, ask for a targeted revision pass.

Revise the draft for clarity and concision. Cut any sentence that restates the previous sentence. Shorten the executive summary to four sentences. Keep technical terms only if the audience needs them; replace others with plain language.

For a proposal template covering scope, timeline, deliverables and stakeholders in a structured format, see project proposal template. ChatGPT for project management covers the broader workflow of running an active project with AI.

How to do this with fenbs

1. Create a free account and add a project for this piece of work.
2. After the proposal is approved, record the confirmed scope, constraints and exclusions as Decisions and rules in fenbs.
3. Add the open questions as tasks in fenbs so they do not get lost in the proposal document.
4. Connect your AI assistant from Integrations so it reads the approved decisions at the start of each session.
5. When follow-on work begins, the assistant already knows what was agreed and what remains open.