Nonprofits come here to figure out how to use AI — and get the tools, data, and skills to do it without building their own. Funders back the commons once, so every grantee has it — instead of each organization rebuilding the same thing from scratch.
Using AI well takes data, infrastructure, and know-how most organizations can't assemble alone. Today there are two bad options: every nonprofit figures it out from scratch, or every foundation builds its own tools for all its grantees and pays to keep them running. The commons is the third option — build the hard part once, and the whole field draws on it. A funder can back it a single time, instead of paying for the same build over and over.
One connected record of organizations, funders, programs, relationships, and evidence — each fact carrying its source and a confidence label. One place to look, instead of a hundred private spreadsheets.
Ready-made tools that do the analysis for you: find the orgs that fit a strategy, map who funds your work, turn a concept into a fundable case. The kind of work that used to need an analyst — as a tool any org can open.
Not just tools, but how to use them well — which sources to trust, what the numbers mean, where AI helps and where it shouldn't. The judgment for using AI responsibly, built in once for everyone instead of relearned org by org.
Nonprofits use it to do the work. Funders back it so their grantees don't have to build it themselves. The networks in between keep it honest and make it better. One tool is live today; the rest are designed, and open as the commons grows.
Proof the model works: the first tool is live. It turns a nonprofit's public IRS filings and public sources into a sequenced, source-grounded funding strategy — the kind of analysis that used to take a research team or a consultant.
The skills are what you see. Underneath, the Commons combines public data, knowledge from the field, and a feedback loop — so a convening or a memo produces an answer and makes the system better for the next person.
Each of these is already built into how the first skill works. The goal is to get this kind of analysis into more hands without putting unsourced or overstated claims into the world.
Insights trace back to a 990 line, a filing, or a public page — never an unsourced assertion. If we can't point to where it came from, it doesn't ship.
The Commons labels what's verified, what's a lead, and what's uncertain. It never dresses a guess as a fact, or stale data as current.
A name on a public list is a lead to verify, not a green light. The tools stop short of outreach and hand off to a person to confirm identity, capacity, and a credible path.
Built on public filings and disclosures. The Commons never invents a donor, a quote, or a relationship to fill a gap — a gap is reported as a gap.
Writeback is opt-in. When the people who know the work push back, that correction updates the graph directly.
The whole point is to put this kind of analysis in the hands of organizations that could never staff a research team — the ones the sector usually leaves out.
Public IRS filings, foundation reports, and academic evidence — resolved into a single connected graph. Grant totals reflect what's traceable in public 990 data, a sample of all giving, and are labeled as such wherever they appear.
Nonprofits use what's live today. Funders and partners decide how fast the rest gets built — back the commons once, and it's there for the whole field, getting better every time someone uses it.
Prepared for the Rustandy Center for Social Sector Innovation and the Center for Applied AI. A collaboration between Emerald South and Azimuth Analysis, with support from an OpenAI grant.