Grant Writing

Draft and refine grant applications with an AI agent that knows your organization.

Overview

The grant writing feature pairs a document editor with an AI agent. Give it your application questions, and it drafts responses using your Reference Materials, memories, and connected integrations as context.

How It's Different from ChatGPT

If you've tried writing grants with ChatGPT, you know the drill: paste your questions, get a draft full of hallucinated statistics, spend an hour cleaning it up, feed it more facts, ask it to cut the word count, realize it introduced new redundancies, repeat. It works, but it takes a long time.

GrantLoop uses a multi-agent approach. When you ask it to write a grant, it delegates to specialized sub-agents that gather information, draft sections, and cross-check the output for accuracy and word count — all in parallel. The goal is to deliver an application that's effectively ready to submit after one prompt, instead of hours of back-and-forth.

You'll see "Task Agent" actions appear in chat when this is happening. The AI decides when to deploy sub-agents based on the complexity of the task.

Creating a Grant Writing Project

  1. Click New at the top of the sidebar
  2. Choose Doc
  3. Name it (e.g., "Ford Foundation 2026" or "State Arts Council LOI")

Or ask the AI in chat: "Create a new grant writing project called 'Smith Foundation Spring 2026'."

Writing Drafts

Ask the AI to draft content. It will search your Reference Materials and connected tools for relevant information before writing.

Basic — Let the agent figure it out:

"Here are the application questions for the Smith Foundation. Please draft responses."

The agent will search through your available data sources — Reference Materials, Google Drive, SharePoint, or whatever you've connected — and find relevant reference material on its own. You don't have to tell it where to look.

Intermediate — Give specific guidance:

"Draft responses to these questions. Include specific statistics and at least one participant story. Reference our 2025 program evaluation for outcome data."

Advanced — Combine with other features:

"Look through our Google Drive for our last two applications to the Smith Foundation. Use those as a style guide. Then draft this year's application, making sure to highlight our new community garden program."

Editing and Revising

After a draft is generated, you can ask the AI to refine specific sections:

  • "Make the second paragraph more concise"
  • "Look through our Google Drive and find that story about the student in our 2024 Riverside Foundation application — the one about overcoming food insecurity. Add something like that to the narrative section."
  • "Rewrite this to emphasize community partnership rather than program expansion"

When the AI proposes edits, you'll see an inline diff showing what will change. Click Accept all to apply or Reject all to discard. Nothing changes in the document until you do — and proposed edits are saved, so navigating away doesn't lose them.

Proposed edits are saved automatically — if you navigate away and come back, your pending edits will still be there waiting for your review.

Word and Character Counts

Grant applications often have strict limits. The AI can measure its own output:

  • "How many words is the project narrative section?"
  • "Count the characters in the first three paragraphs"
  • "This section has a 500-word limit. Is it under?"

The agent verifies counts before finalizing drafts, so you're less likely to run into surprises.

Creating Budgets

The AI can generate Excel budget spreadsheets with working formulas, input tabs, and matching fund calculations. It learns your organization's budget style by searching through your Reference Materials and connected integrations.

"Make me a $50,000 budget for a community gardens program, please."

For the full guide — including how to use matching funds, pull from integrations, and get the formatting right — see Budgets.

Best Practices

  1. Upload reference materials first — The AI writes dramatically better grants with 100+ pages of context
  2. Teach it with Memory — Set up standing instructions for your writing preferences
  3. Iterate — Ask for revisions, provide feedback, refine until it sounds like you