AI4Good Foundation is opening applications for our Fall 2026 pilot cohort. Over the next six weeks we will select a small number of institutional partners to co-launch one of our three program pillars: Career Navigator, Skills Translator, or the Educator AI Toolkit. Pilots run for six months, cost the partner institution nothing, and include an independent third-party evaluation report that the partner can use for its own reporting, accreditation, and board communication.
We are publishing this call early and in plain language because we want partners who self-select on mission alignment, not on the flash of a press release. If your institution serves underrepresented learners and you have been waiting for a way to engage with open-source AI on terms that respect your community, this is that invitation.
Who qualifies
We are looking for institutional partners whose primary mission is to serve learners in underrepresented communities. Concretely, that includes:
- Community colleges and the state-level community college systems that coordinate them.
- Public workforce development boards and the American Job Centers they operate.
- Refugee resettlement agencies and the larger national networks that support them.
- Title I high schools and the districts that house them.
- Minority-serving institutions, including Historically Black Colleges and Universities, Hispanic-Serving Institutions, and Tribal Colleges and Universities.
- Community-based organizations whose programming includes adult education, credential navigation, or workforce reentry for underrepresented learners.
We are not the right partner for selective four-year universities, enterprise employers, or commercial training providers, even when those organizations sincerely serve some underrepresented learners. Our resources are limited and they go to institutions where the gap between the AI access learners need and the AI access they currently get is largest.
What AI4Good provides
Every pilot, regardless of pillar, includes the following at no cost to the partner institution.
- A working, configurable deployment of the program tool, built on open-weight foundation models, with deployment documentation written for institutional IT staff rather than research engineers.
- Onboarding and configuration support, including a working session to adapt the tool to the partner's local context: local credentials, regional labor-market focus, language coverage, intake workflow.
- A pre-registered pilot design and evaluation plan, developed jointly with the partner and reviewed by an independent third party. Pilot designs prioritize outcomes that the partner institution actually cares about (credential completion, job placement, advisor time-on-task) over engagement metrics.
- The third-party evaluation report itself, delivered at the end of the pilot, with separate versions tuned for accreditation, board, and public audiences.
- Practitioner office hours throughout the pilot, so front-line advisors, case managers, and instructors have a direct channel for questions, edge cases, and improvement requests.
What we ask of partners
A pilot is a real commitment from the partner institution. The ask is modest, and we are explicit about it up front.
- A named institutional point of contact who can coordinate across the program staff, the IT environment, and the evaluation work.
- A small group of front-line practitioners (advisors, case managers, instructors) who will use the tool with learners and participate in monthly feedback sessions.
- Cooperation with the third-party evaluator on a pre-registered evaluation design. The evaluator handles instrument design, data collection, and analysis; the partner provides access and context.
- Willingness to be publicly named as a pilot partner once the partnership is finalized, and to be named in the published evaluation report. Confidential pilots are possible in exceptional cases but they are the exception.
What we will not do
A handful of commitments are worth stating plainly, because they are how we want this field to work.
- We do not charge for pilots and we do not charge for the tools they pilot.
- We do not ask for exclusivity. Partner institutions are free to adopt any other tool, open or commercial, alongside ours.
- We do not sell learner data and we do not share learner data with any third party. Period. Data handling is documented in plain language and translated for learner consent.
- We publish evaluation results openly under permissive licenses, including null and negative results. If a tool we ship does not work for your population, we want the field to know that, too.
How to apply
Send a short note (a few paragraphs is plenty) to contact@thefoundationai.org introducing your institution and the learners you serve, naming which program pillar feels most aligned with your current needs, and identifying a point of contact we can write back to. There is no application form to wrestle with. We will reply within two weeks with next steps, which for shortlisted partners is a thirty-minute call with our programs team. For more on the programs themselves, see the programs page and the partners page.
We are building this for the long term. The institutions that join the first cohort will shape how the work is done for every cohort that follows.