Open fundraising database
Find the funders who already fund work like yours.
Search every U.S. foundation by what it has actually funded, straight from public IRS filings. Verified facts stay verified. Anything our AI adds is labelled.
Free. No account needed. Search never uses AI credits.
- Verified from IRS filings
- AI suggestions always labelled
- Open source, open data
Live from the database.Database refreshed Sources and known limits
How it works
Three steps from a sentence to a shortlist.
Step 1: Describe your work
Type a sentence about what you do and where. Search reads what funders actually paid for, not just their names.
Step 2: Read the filing, not a summary
Each profile shows what the funder reported: how to apply, money by year, grants paid, officers. Every fact links to its source.
Step 3: Save it and act
Create a free account to save funders, run a pipeline, keep tasks, and ask the AI to explain fit or polish a draft. AI output stays labelled.
Three kinds of information
You always know what you are reading.
Every value on a funder page is one of three kinds. Each kind looks different, and colour is never the only signal.
Source
Verified from a public filing.
Teal, with a dotted underline and a document icon. Click it to see the filing, the year and, where we have it, the file fingerprint.
Example
- Grants paid
- Total revenue
- Not available
- Applications
- Accepts applications
- Or
A missing line reads “Not available.” An absent statement reads “Not stated in filings.”
AI
Suggested by a model. Never a fact.
Violet, with a dashed border and the word “AI.” Every point links to its evidence. You can accept, edit or dismiss it.
Example
Fit analysis
Example: recent grants went to food programs in the same county as your organization [E1]. Typical grant size is close to your ask [E2].
Accept, edit or dismiss. Your verdict is recorded.
The AI badge appears on every machine-suggested value, in the app and in exports.
Yours
Your own notes, stages and tags.
Green, with a solid left rule. Only you and your team can see or change it.
Example
Your notes
Example: met the program officer at the regional conference. Next action: send a two-page summary.
Stages, tiers, owners, tasks and notes live here. Nothing in this class ever reads as a filing fact.
Honest by design
Seven rules every page keeps.
These are written into the project's governance. A change that breaks one is not merged.
Unknown is not closed.
When a filing does not say whether a funder accepts applications, we say “Not stated in filings.” We never say “closed.”
Missing is not zero.
A number that is not on the filing reads “Not available.” It never reads “$0.” A real zero reads “$0.”
Contacts are opt-in and never bought.
We show a contact only when the public record allows it. Role inboxes like grants@ can appear. A named person’s address does not. We never use vendor lists.
Every fact has a source.
Each value carries the dataset, the filing year and, where we have it, a fingerprint of the exact file we read. Click it and check.
AI is labelled and cites its evidence.
Anything a model writes is marked AI, points to the evidence it used, and never claims a funder is interested in you.
No made-up data.
No invented rows. No demo data in the real database. If we do not have it, we say so.
Amended filings replace the original.
When a funder files an amendment, the newer return wins everywhere. Nothing is counted twice.
The full list, and how decisions are made, is on the open source page.
Free and paid
Search is free for everyone. Credits pay for the AI.
Only calls to a language model cost credits. Every plan includes every search and every profile.
Free
$0
- Unlimited funder search and full profiles: where every fact came from, and whether the funder says it accepts applications
- 25 AI credits a month: about 5 fit analyses or 2 web research reports
- 50 saved funders and one pipeline
- Outreach drafts (uses credits), 100-row CSV export
Paid plans
from $10 / month
- 150 to 25,000 AI credits a month, more seats, full CSV export
- Pro ($20): send outreach through your own Gmail, each message approved by you
- Team ($100): a knowledge base shared across the team, follow-ups that stop when a funder replies, and the API
- Paid use funds the free tier and the open-source project
Open source, self-install
Run the whole thing yourself.
Apache-2.0 code, CC BY 4.0 data, no plan limits. Three parts: build the database, create the app role, run the web app.
cd corpus
uv sync
cp .env.example .env # set DATABASE_URL and SEC_USER_AGENT
uv run funderdb doctor # checks your setup
uv run funderdb migrate # creates the schema
uv run funderdb bootstrap --profile small # a laptop-sized slice, under an hourcreate role getfunded_login login password '<choose a password>' in role getfunded_app;cd apps/web
npm ci
cp .env.example .env.local # DATABASE_URL (getfunded_login), Supabase Auth keys, SELF_HOSTED=true
npm run db:migrate # creates the getfunded schema
npm run db:ping # proves the app can read the corpus and cannot write it
npm run dev # http://localhost:3050Built for AI agents
Written so a coding agent can install, run and extend it.
Each part of the repository has an AGENTS.md with the rules that matter: data honesty, the read-only corpus boundary, metering, no private data. Team plans add an API.
Read README.md, docs/ARCHITECTURE.md and apps/web/AGENTS.md. Then install GetFunded locally using corpus/docs/SELF-INSTALL.md with the small profile and tell me what you found.Start with a search. Sign up when you want to keep what you find.
No credit card. Name and email only. The free plan includes every search and a monthly allowance of AI credits.