Grant writer copilot: Better proposals in half the timeCopy
This AI workflow transforms your grant writing process by systematically analyzing RFPs, generating tailored proposal drafts, integrating evidence, and refining content to match funder preferences – turning what typically takes days into a focused 2-3 hour session that produces a compelling, submission-ready proposal.
Benefits
- Cut proposal writing time by 60-70%, from days to hours for most standard grants
- Never miss a requirement with AI-powered RFP analysis and automated checklist creation
- Improve proposal quality and consistency across all submissions with AI-guided structure and tone
- Increase win rates by ensuring perfect alignment with funder priorities and language
- Reduce writer’s block and stress with instant first drafts you can refine rather than starting from scratch
Resources
⏰ Time:
2-3 hours for a standard proposal (may be longer for complex grants or first-time applications)
👤 People:
1 primary writer familiar with the program being funded, plus optional reviewers. Ideally someone who knows your organization’s programs, impact data, and has basic grant writing experience.
🛠️ Tools:
- ChatGPT
OPTIONAL:
- Grammarly or similar for final proofreading
- Canva or similar for creating simple visuals/infographics
- Google Docs/Word for collaborative editing if working with a team
📂 Internal data:
- Grant RFP/guidelines document or webpage
- Organization’s basic information (mission, programs, budget)
- Program description for the grant you’re applying for
- Basic impact data or outcomes
OPTIONAL:
- Previous successful proposals (as templates/reference)
- Detailed budget breakdown
- Letters of support
- Evaluation plans or logic models
- Staff bios and organizational history
Workflow
This entire workflow is designed to be completed in a single ChatGPT conversation, allowing the AI to build comprehensive understanding of your proposal needs.
ℹ️ Note
This workflow is focused on writing a grant proposal based on an RFP document, but if you need to create a different document (e.g. LOI) or don’t have a RFP, you can still use this workflow with a few adjustments. Just give all the context you have (documents, emails, URLs…) and explain what you need exactly.
STEP 1: Grant RFP analysis
Begin by having AI analyze the grant requirements to create a comprehensive checklist and outline. This ensures you don’t miss any requirements and understand exactly what the funder is looking for.
Start a new ChatGPT conversation and upload the RFP document (or provide the URL/text):
Act as an expert grant writer specializing in nonprofit funding.
I need you to analyze this grant RFP/guidelines and create a comprehensive proposal outline.
# DATA
**UPLOAD RFP FILE, MENTION URL OR PASTE TEXT**
# REQUEST
Please provide:
1. ELIGIBILITY CONFIRMATION
- List all eligibility requirements
- Flag any potential concerns based on common nonprofit limitations
2. REQUIRED SECTIONS & COMPONENTS
- List every required section in order
- Note word/character limits for each section
- Identify all required attachments/supporting documents
3. EVALUATION CRITERIA
- Extract the exact scoring rubric or evaluation criteria
- Note any stated preferences or priorities
4. KEY THEMES & LANGUAGE
- Identify recurring keywords and phrases the funder uses
- Note any specific terminology we should mirror
5. CRITICAL DETAILS
- Application deadline and submission method
- Budget restrictions or requirements
- Reporting requirements if funded
- Any unique or unusual requirements
Format this as a clear checklist I can work through systematically.
STEP 2: AI-Generated first draft
Now we’ll generate the first draft of your proposal. We’ll break this into manageable sections rather than trying to create everything at once.
First, provide context about your organization and program (continue in the same ChatGPT conversation):
Now I'll provide information about our organization and the program we're requesting funding for. Use this context for drafting the proposal.
# DATA
ORGANIZATION OVERVIEW:
- Organization name: **NAME**
- Mission statement: **MISSION**
- Years in operation: **YEARS**
- Annual budget: **BUDGET**
- Geographic area served: **LOCATION**
- Target population: **POPULATION**
PROGRAM TO BE FUNDED:
- Program name: **PROGRAM NAME**
- Brief description: **DESCRIPTION**
- Problem it addresses: **PROBLEM STATEMENT**
- Target beneficiaries: **WHO WILL BENEFIT**
- Key activities: **MAIN ACTIVITIES**
- Expected outcomes: **OUTCOMES**
- Program budget: **AMOUNT REQUESTING**
- Timeline: **DURATION**
IMPACT DATA:
- Outcomes: **List your specific metrics, outcomes, success stories**
- Problem: **Include relevant statistics about the problem**
- Studies: **Add any research or studies that support your approach**
# REQUEST
Based on the RFP requirements and the information provided above, write the first section required in the RFP.
WRITING GUIDELINES:
- Start with a compelling hook that addresses the funder's main priority
- Clearly state the amount requested and program name
- Include 2-3 sentences on our credibility/track record
- End with anticipated impact that aligns with funder goals
- Mirror the funder's language and keywords identified earlier
Write in active voice, avoiding jargon unless the funder uses it.
# OUTPUT
From now on, when I write "next", give me the text for the next section required in the RFP (follow the order specified in the RFP).
I may ask for improvements or extra information before moving to the next section.
If the RFP asks for other documents (e.g. cover letter, executive summary, appendixes...) help me create them too. Give me the full text for those documents (preferred if possible) or at least recommendations to create them.
💡 Pro tip
You can upload relevant documents or provide URLs (instead of filling manually the DATA section or to provide additional context). If you have good documents (program briefs, impact reports, detailed “About Us” and “Programs” pages on your website…), you can probably improve results and save time by uploading them.
If your documents are very long or have many pages that are not relevant for this grant, you might want to ask first for a summary of the relevant parts (using ChatGPT or maybe NotebookLM) and feed only the summary to this prompt. Providing lots of long documents may confuse the AI or go over its maximum context window.
If you are not happy with the results, feel free to ask for changes. When you think the results are OK, write “next” and ChatGPT should give you the text for the next section.
STEP 3: Incorporate data and evidence
Now we’ll strengthen the proposal with specific evidence and data. This step ensures your claims are backed by concrete proof.
I need to strengthen the proposal with specific evidence. Here's the data and evidence we have available:
# DATA
IMPACT DATA:
**Copy or upload success metrics, relevant statistics, studies, etc.**
QUOTES/TESTIMONIALS:
**Copy or upload quotes from beneficiaries, partners, staff....**
# REQUEST
Please revise the relevant sections to incorporate this evidence smoothly. For each data point:
- Integrate it naturally into the narrative
- Ensure it directly supports our case
- Use it to strengthen our credibility
Show me the revised portions with the data integrated, highlighting what changed.
If you think some of the data provided above is not relevant or should not be incorporated to this proposal, don't add it and explain why at the end.
If you don’t have specific data yet, use this alternative:
I need to identify where we should add supporting evidence. Please review the draft and:
1. Mark every claim that needs data support with [EVIDENCE NEEDED: type of data]
2. Suggest the type of evidence that would be most compelling for each claim
3. Provide example statistics or benchmarks from similar programs if you know them
4. Indicate which data points are critical vs. nice-to-have
This will help me gather the right evidence before submission.
STEP 4: Add supporting visuals or media
Visuals can significantly strengthen your proposal if the format allows. Let’s identify opportunities and create simple graphics if needed.
Review our proposal draft and identify where visuals could enhance our application.
For each opportunity, specify:
1. LOCATION: Which section would benefit
2. TYPE: What kind of visual (chart, infographic, photo, logic model, etc.)
3. PURPOSE: What it would communicate
4. PRIORITY: Essential, helpful, or optional
If the RFP specifically mentions or encourages visuals, note those requirements.
For any essential visuals, get AI help creating them:
Help me create the visual resource mentioned below:
# DATA
VISUAL DESCRIPTION
**Type of visual, context, goal...**
DATA/CONTENT FOR VISUAL:
**Provide the data or content to include**
RESOURCES AVAILABLE:
**Mention the tools and people/skills that you have to create your visuals**
# REQUEST
If you can create the visual directly, do it now.
Otherwise, give me detailed recommendations to create it (including design suggestions to make it compelling and clear, which tools and features to use, specific text to include, maybe prompts to use on other AI tools, etc.).💡 Pro tip
ChatGPT and some other AI tools are already multimodal, meaning they can create images directly in the normal chat. But they still have limitations (size & format, text, etc.), so sometimes it’s better to use AI tools that are optimized only for images (e.g. Midjourney, Recraft, Leonardo…) or “traditional” design tools with AI features (Canva, Adobe Suite, etc.). Still, you can use ChatGPT to give you ideas based on all the previous context it has. Maybe even prompts for those other tools.
STEP 5: Comprehensive Review
We will now simulate a ‘Red Team’ review, where the AI’s sole job is to find every possible weakness, inconsistency, and missed opportunity before submission. This prompt is deliberately harsh to catch everything.
This is the final review. I need you to act as a highly critical, expert grant reviewer who is trying to find reasons to reject this proposal. Your goal is to identify every single weakness so I can fix it.
# REQUEST
Please perform a comprehensive "Red Team" review of the entire grant package (main narrative, executive summary, budget narrative, etc.) and provide a detailed report. I need you to check the following:
RFP COMPLIANCE AUDIT:
- Create a checklist of every single requirement (e.g., every required section, attachment, word count, font size, formatting rule).
- Confirm if our draft package 'Passes' or 'Fails' each item. Be ruthless. If a word count limit is 500 and a section is 510, that is a 'Fail'.
EVALUATION CRITERIA ALIGNMENT:
- Quote the 3-5 sentences from our draft that are the weakest or least persuasive in aligning with those criteria.
- For each weak sentence, explain *why* it fails and suggest a specific, stronger revision.
INCONSISTENCY REPORT:
- Scan all documents for any contradictions or inconsistencies.
- Pay close attention to numbers (e.g., budget total vs. amount mentioned in the narrative), project goals (e.g., executive summary vs. main proposal), and beneficiary numbers.
REPETITION & JARGON AUDIT:
- Identify any overused 'buzzwords' (e.g., 'innovative,' 'synergy,' 'impactful') and repetitive phrases that make the proposal sound generic.
- List them and suggest 3-5 stronger alternatives for each.
"VOICE & HEART" REVIEW:
- Does the proposal have a compelling, confident, and passionate voice, or does it sound robotic and AI-generated?
- Identify any sections that feel 'cold' or 'generic' and suggest how to inject more storytelling or human-centric language.
FINAL READINESS SCORE:
- Finally, provide a single 'Submission Readiness Score' from 1-10 (1=Not ready, 10=Flawless).
- Write one final paragraph summarizing the single most critical revision I need to make before I can confidently submit this application.
💡 Pro tip: Get a “Second Opinion”
Once you have a strong final draft from your primary AI, consider getting a “second opinion”. Copy the entire proposal and paste it into a new conversation with a different AI model (for example, if you used ChatGPT, try Claude or Google Gemini).
Different AI models have unique training data, settings and strengths. One model might be exceptional at catching subtle grammatical errors, while another might be better at identifying data inconsistencies or suggesting more persuasive language. This “fresh set of eyes” can catch blind spots that you and your original AI assistant might have developed during the drafting process.
You can use the prompt above also with this new AI. But you need to give context to this new AI, so you should add something like this to the prompt:
# DATA
I've uploaded a file with the grant guidelines/RFP and uploaded **(or copied below)** the draft proposal that you have to review.
**Upload the original RFP file and upload or copy your draft proposal**Recommendations
- Provide all relevant context. Before starting, gather all documents: RFP, program descriptions, impact data, budget info, etc. Sometimes people complain about AI giving very generic outputs, but in many cases it’s because you are not giving the AI enough context to provide a good personalized output. The more specific context you provide, the better the output.
- Reuse context and conversations. If you have to start from scratch for every proposal, you will not be very efficient. Instead, save your AI conversations and final proposals. Next time you apply for a similar grant, you can start with: “Here’s a successful proposal we submitted previously. Let’s adapt it for this new opportunity…”
- Don’t trust ChatGPT (or other LLMs) with numbers. Always double-check any calculations, budget figures, or statistical claims the AI makes. Use AI for narrative and structure, but verify every number yourself. If you need complex calculations, you can ask ChatGPT to create the code for a Python calculator, or maybe give you the formula or script for Google Sheets or Excel.
- Watch for repetition. AI tends to repeat phrases and ideas. After generating all sections, do a final read to eliminate redundancy. Ask the AI to help: “Identify any repeated phrases or ideas across sections and suggest alternatives”.
- Use human review for heart. AI excels at structure and incorporating requirements, but may miss the emotional resonance. Have a human reviewer focus on: Does this tell our story compellingly? Will it move the reader to act?