Most of grant writing is reading carefully and structuring an argument, and that is where Claude helps most. Use it to turn a funder's guidelines into a checklist, to sharpen a needs statement around your own data, to build the logic model that links problem to outcome, to justify every budget line, and to draft the report you owe at the end. Do not use it to find facts about funders or your community: it will invent them. Your data in, a tighter proposal out.
What should Claude never do in a grant?
Start with the limits, because a single invented fact can sink an application and a relationship with a funder.
- Invent statistics. Community need figures must come from your own records or a named public source you can cite.
- Research funders from memory. Priorities, deadlines and award sizes change every cycle. Read the current guidelines and paste them in.
- Write a letter of support as if from a partner without that partner reading and signing it.
- Submit anything you have not read aloud. Reviewers notice a generic voice, and some funders now ask how AI was used.
Inside those limits, it saves days.
Step 1: turn the guidelines into a checklist
The quote-everything rule makes the checklist checkable. If Claude cannot point to the line, the requirement does not exist. That alone catches the most common reason applications are rejected before anyone reads them: a missing attachment or a blown page limit.
Parses what a funder actually requires, builds the logic model from problem to intervention to outcomes, drafts a budget narrative that justifies each line, and helps with letters of support. You bring the data and the program; it holds the structure.
Get the AI Grant Writer →Step 2: a needs statement built on your data
Step 3: the logic model and the budget narrative
Funders want to see a straight line from activity to outcome. Ask Claude to draw it as a table: inputs, activities, outputs, short-term outcomes, long-term outcomes, and for each outcome how you will measure it. Then ask the harder question: which link in the chain is weakest, and what evidence would a skeptical reviewer want?
Orphan lines are where reviewers stop trusting a budget. Finding them before submission is worth more than any amount of polishing the narrative.
Need more than one specialist?
Buying the AI Grant Writer gets you one file. The Library subscription loads every skill in the catalog through one connector, in Claude, ChatGPT Plus and up, Claude Code, Codex or Cursor, and keeps each skill updated when we improve it. $9 a month or $59 a year, with a free trial, no card needed.
Start the free trial → Why a subscription beats buying filesWhich funders should you apply to at all?
The fastest way to raise your success rate is to stop applying to funders who do not give in your space. Claude cannot look up who funds what, but it can read what you collect: a foundation's published guidelines, its annual report, its list of past grantees. Paste those in and ask for a fit score against your program, with the evidence quoted. For a strategy across many funders, the next skill is built for it.
Matches your program to funders who actually give in your space, then writes the ask. Built for community, family and corporate foundations and local government grants, not federal research funding.
Get Marisol →Step 4: the report you owe the funder
Honest reports renew grants. Program officers know that plans change; what they do not forgive is finding out from someone else.
Where does Claude struggle with grants?
- It does not know current funder priorities. Always paste the latest guidelines.
- Its default voice is generic. Give it two paragraphs your organisation has written and ask it to match them.
- It overpromises outcomes. Ask for targets you can measure with data you already collect.
- Federal and research grants have their own rules. Agency formats and compliance requirements are strict; use the official instructions as the source of truth.
- It cannot build the relationship. A call with the program officer before you apply still matters more than any draft.
How do you install a skill in Claude?
A KissMySkills skill is a ZIP file with a SKILL.md at its root. In the Claude desktop app, open Customize → Skills, add a skill, upload the ZIP and toggle it on. Skills need code execution, which lives under Settings → Capabilities; on Team and Enterprise plans an owner controls it. Anthropic documents the steps in its Skills help article, and our step-by-step install guide has screenshots.
In ChatGPT or Gemini, open the SKILL.md in a text editor and paste it into custom instructions: a Claude Project, a custom GPT or a Gem. With a Library subscription you skip the file entirely: connect KissMySkills once and ask for the specialist by name in the chat you already use.
In summary:
Use Claude for reading and structure: the checklist, the needs statement, the logic model, the budget narrative and the report. Never let it supply facts. The AI Grant Writer ($19) covers single applications; Marisol ($19) covers funder strategy across a pipeline.
AI Grant Writer
Requirements checklist, logic model, budget narrative and letters of support, built from your data and the funder's own words.
Or get it with every other skill in Library: $9/month, free trial, no card needed.
KissMySkills is a marketplace of 2,500+ AI skills, agents, prompt packs & free tools for Claude, ChatGPT & any AI chat.
Frequently asked questions
Can I use Claude to write a grant proposal?
Yes, for structure and drafting: checklists from the guidelines, needs statements from your data, logic models, budget justifications and reports. You must supply every fact and read the final text yourself.
Do funders accept AI-written grant applications?
Policies vary. Some funders say nothing, some ask you to disclose AI use, and some research funders restrict it. Check the current guidelines of each funder and be ready to explain how you used it.
Can Claude find grants for my nonprofit?
Not reliably from memory, because funder priorities and deadlines change every cycle. Find candidates in a grants database or funder websites, then paste their guidelines into Claude to score the fit.
How do I stop Claude making up statistics in a grant?
Give it only your own data with sources and tell it to mark any claim that needs a number you have not supplied, instead of filling the gap.
What is a logic model in a grant proposal?
A table that links inputs and activities to outputs and short and long term outcomes, with how each outcome will be measured. Claude is good at drafting one from your program description and finding the weakest link.
Related guides
- How to Use Claude for PowerPoint Presentations
- How to Install a Claude Skill
- AI Agents vs Skills vs Prompts: What Is the Difference?
