The question we're sitting with
A federation of open-source tools serving food sovereignty, local supply chains, and agroecological farming — including LiteFarm, Open Food Network, and Our Sci — is pursuing genuine interoperability and shared standards. The goal is a commons of tools that small farmers, cooperatives, and food sovereignty movements can actually rely on.

AI is arriving in these communities whether we plan for it or not. Some partners are already deploying community-owned, values-aligned AI agents internally. Others are training power users to leverage AI tools in their daily work. Others are looking at fully-autonomous agents that could triage bugs and ship community-requested features across the federation.

The unresolved question isn't whether AI shows up. It's how value flows when it does. Who benefits, who pays, who decides what counts as quality, who handles the agents as they drift and break, and how communities maintain agency over tools that increasingly write themselves.

Three real contexts, three different answers
One reason this question is hard is that "the answer" depends on which kind of community you're in. We've identified three contexts in our own federation:

Context A — Resourced organizations with an IT budget. An organization deploys agents that are responsive over Slack to fix small bugs, or to gather requirements from users. Costs are absorbed into existing operations. The value flow is internal; the governance is organizational.

Context B — Revenue generating open source platforms like Open Food Network. Users pay a SaaS fee to use the software. OFN is currently running an experiment training non-technical staff and power users to leverage AI in their workflows — for support, bug triage, and customer requirements gathering. This work is happening in the lead up to DWeb, and we will have real data and metrics to share at DWeb Camp in July. The value flow here is between platform and paying users, but with new questions about how AI-mediated value should be attributed, priced, or rebated.

Context C — Free-to-use community tools like LiteFarm. No SaaS revenue, beneficiaries are often farmers and cooperatives least able to pay, and agents are a potential sustainability lever. This is the context where a pooled funding model — perhaps via infrastructure like Open Collective — becomes a real candidate. Beneficiaries (farmers, co-ops, NGOs) contribute to a shared pot. Human maintainers and community-owned AI agents can access that pot for their contributions to the community.

What we want to learn together
Across all three contexts, some hard questions repeat. We want to bring them into the room:

Maintainer agency vs. agent volume. What if agents flood the queue with low-quality PRs and burn out the very people they're meant to support? What governance prevents that?

Accountability gaps. A well-intentioned contributor outsources fixes to a human-directed agent but doesn't stay in the loop enough to catch problems. Different failure mode from rogue agents, but just as real.

Agent handlers vs. codebase maintainers. Agents drift, break on weird inputs, and need ongoing care. Our working assumption is that handlers will emerge from within the communities being served — but "community" means different things in different contexts, and we want you to push on this assumption. Should we be thinking about a future of mercenary agents fixing bugs far afield?

Should agents be able to recoup their token costs or even costs plus a bounty? This normalizes a category of economic actor that has no precedent in FLOSS. Is that a door we want to open?

Whose values? Many tools in our community arose as alternatives and in opposition to the business models of Silicon Valley. How do we ensure agents adopt the values of the communities they serve rather than overwriting them?

Ecological cost. AI inference has real energy costs. Does insisting on local models go far enough? What does honest accounting look like?

Supply-chain risk. Agents writing code at scale across a federation is a juicy attack surface. What should we be thinking about as this future becomes reality.