BRENNEN

Industries / Government, education and community / Nonprofits / Foundations

Working with other funders

Foundations pool money, share checks on grantees and fund together, through community foundations, alliances and pooled funds.

5

Milestones achieved, from 1914 to 2024.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1914

    Many donors, one board

    Frederick H. Goff, president of the Cleveland Trust Company, set up the Cleveland Foundation on 2 January 1914 to unite "gifts of various sizes left at different times by different donors" under a "living hand" that could change how the money was used as needs changed, rather than the "dead hand" of fixed wills. The Cleveland Foundation calls itself the world's first community foundation.

    case.educlevelandfoundation.org

    Credit

    Frederick H. Goff

  2. 2000

    A vaccine alliance

    The Gates Foundation's initial $750 million commitment in 2000 was, by Gavi's account, "catalytic in establishing Gavi" and in drawing support from other donors to the vaccine alliance. The foundation has pledged again for each period since, most recently $1.6 billion for 2026 to 2030.

    gavi.org

    Credit

    Bill & Melinda Gates Foundation; Gavi, the Vaccine Alliance

  3. 2013

    One check, many funders

    A US foundation that gives to a foreign charity can make an equivalency determination, a finding that the charity is the equivalent of a US public charity, and before NGOsource one foundation could not rely on another's. NGOsource, a project of the Council on Foundations and TechSoup, has kept a shared store of these determinations since March 2013 that members can reuse for a lower fee. By its own count, in its first ten years its members gave more than $8 billion in unrestricted funds to more than 8,000 organizations in 161 countries.

    nonprofitlawblog.com

    Credit

    Council on Foundations; TechSoup

  4. 2022

    An alliance that missed

    The Alliance for a Green Revolution in Africa (AGRA), launched in 2006 and two-thirds funded by the Gates Foundation, with the Rockefeller Foundation among its lead donors, first aimed to double incomes for 20 million farming households and halve food insecurity in 20 countries by 2020. An evaluation by Mathematica, released in February 2022 and paid for by the Gates Foundation on behalf of the lead donors, found that "AGRA did not meet its headline goal of increased incomes and food security for 9 million smallholders"; maize yields rose in Ethiopia, Ghana and Nigeria but not in Tanzania, Burkina Faso or Kenya.

    LessonResearchers outside AGRA had reported the shortfall in 2020, two years before the donors' own evaluation, after AGRA declined to give them its internal monitoring data.

    usrtk.orgscidev.net

    Credit

    Gates Foundation, Rockefeller Foundation and government donors (AGRA's funders); Mathematica (evaluation)

  5. 2024

    Pooled funds counted

    The Bridgespan Group surveyed collaboratives, funds that pool or channel money from many donors, from 2021 to 2023, and 280 responded at least once. Together they deploy $4 billion to $7 billion a year, and the 35 that each deploy more than $25 million a year account for over 75% of the money.

    LessonPooled giving is concentrated in a few large funds.

    bridgespan.org

    Credit

    The Bridgespan Group

Watch for

  1. Ten funders on AI

    Ten foundations led by the MacArthur Foundation and Omidyar Network launched Humanity AI in October 2025, committing $500 million over five years to work on AI and the public good, with a pooled fund managed by Rockefeller Philanthropy Advisors. In May 2026 it announced more than $18 million in new grants, including $500,000 each to groups such as the AI Now Institute and the Pulitzer Center, and a $10 million open call.

    Achieved whenGrants paid by the coalition and its pooled fund reach the pledged $500 million, shown in public reports or tax returns.

    adn.commellon.org

    Claimed by

    The Humanity AI coalition, October 2025 and May 2026

  2. One donor becomes many

    Open Philanthropy, which had worked mainly with one funder, Good Ventures, renamed itself Coefficient Giving in November 2025 and turned its focus areas into funds that other donors can join, saying it is "focused on growing our work with funders outside of Good Ventures".

    Achieved whenIts published figures show a large and rising share of its grants paid with money from donors other than Good Ventures.

    forum.effectivealtruism.org

    Claimed by

    Coefficient Giving, November 2025

Open

  1. 01

    One due diligence pack

    A shared due diligence pack that a charity fills in once, covering its governing documents, accounts, board, policies and registration checks, in an open data format that several funders agree to accept in place of their own checks, tested with five UK or US funders in one grant round.

    The brief
    Credit · unclaimed
  2. 02

    One allocator or many

    A simulation on real UK grants data that compares two designs: each of 20 funders' AI agents allocating its own budget alone, and one shared agent allocating all 20 budgets to the funders' published priorities. Measure, for each design, how many charities are left with no funder after one simulated funder withdraws, and what share of money goes to charities that already have four or more funders.

    The brief
    Credit · unclaimed

Open 01 · the brief

One due diligence pack

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A shared due diligence pack that a charity fills in once, covering its governing documents, accounts, board, policies and registration checks, in an open data format that several funders agree to accept in place of their own checks, tested with five UK or US funders in one grant round.

Why it is still open

Shared checks exist for single tasks. NGOsource lets US funders reuse equivalency determinations, and in humanitarian aid, funders such as the UK's Foreign, Commonwealth and Development Office draw on audits by the Humanitarian Quality Assurance Initiative (HQAI), a practice called passporting. IVAR wrote in August 2026 that "a shared, funder-agnostic profile that organisations complete once and any funder can draw on" needs no AI, and that for shared repositories "the barrier here isn't the technology, it's adoption". Project Streamline recommended shared data repositories in 2008.

nonprofitlawblog.comhqai.orgivar.org.ukatlanticphilanthropies.org

Built with what exists

JSON Schema to define the pack's format; org-id.guide identifiers, the scheme 360Giving uses to link organizations to their official registers; the public data services of the Charity Commission for England and Wales and of Companies House to fill in details; an AI model to map each funder's existing checklist onto the pack.

  1. First step

    Collect the due diligence checklists of 20 funders and map which items they share.

  2. Done looks like

    Five funders accept the pack for one round and at least 30 charities use it; applicants' hours on due diligence measured against the same funders' previous round; funders report any check the pack missed.

Open 02 · the brief

One allocator or many

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A simulation on real UK grants data that compares two designs: each of 20 funders' AI agents allocating its own budget alone, and one shared agent allocating all 20 budgets to the funders' published priorities. Measure, for each design, how many charities are left with no funder after one simulated funder withdraws, and what share of money goes to charities that already have four or more funders.

Why it is still open

The 2018 Global Philanthropy Report found that 58% of the foundations it identified do not collaborate with other foundations. 360Giving's 2019 analysis of London funders found 68 recipients with yearly income of £100,000 to £1 million that shared four or five funders, while only a handful of smaller groups did. In Anthropic Fellows' simulations where people shared one resource, one agent serving everyone beat one agent per person in all four settings tested (see the Note "Many agents, one budget"), but this research found no such test on grant money.

ubs.com360giving.org

Built with what exists

GrantNav downloads from 360Giving; AI agents built on Claude or another model through its programming interface, each given a funder's published priorities; Python for the simulation and network measures.

  1. First step

    Download one year of GrantNav data for 20 London funders and reproduce 360Giving's count of shared recipients.

  2. Done looks like

    Three years of grants from 20 funders replayed under both designs, with both measures reported; prompts and code published; a panel of funders reviews whether the simulated choices fit their stated priorities.