Building an AI Commons for Brussels NGOs
A hands-on working session for NGO people, techies, and commons practitioners to design a shared AI infrastructure that Brussels' civil society can own and govern together — instead of renting it from Big Tech. First living-lab session of the Architecting Digital Sovereignty series, at Commons Hub Brussels.
## purpose of the workshop
Brussels NGOs are being pushed to "adopt AI" — but almost every option on the table means handing data, budget, and dependency to the same handful of American platforms. We think there's another way.
This workshop brings together NGO people, techies, and commons practitioners to ask a different question: what would it look like if Brussels' civil society owned its AI infrastructure together, instead of renting it? Not a pitch, not a product demo — a working session to start designing a shared AI resource that NGOs govern collectively, the way communities have governed shared resources for centuries.
This is the first of a short series of living-lab sessions where we build the thing as we learn how to govern it. Along the way, we want participants to actually grasp the issues underlying a digital commons — what makes it different from a shared tool or a free service, and what it demands from the people who keep it alive.
## Target group
We're deliberately bringing two different roles into the same room — the people who would use the commons and the people who would build and maintain it — because a commons only works when both sides negotiate it together, from day one.
**Users**
- **NGO IT staff** (IT, comms, data, or leadership) curious about AI but wary of the proprietary-platform trap — the people who would actually run on this shared resource
**Contributors**
- **Techies & IT people** who know how to set up a self-hosted LLM on our own premises — the people who would build and keep the resource running
- **Commons & governance practitioners** interested in a live case study, not just theory — the people who help shape how it's governed
- Anyone from the Brussels civil-society/tech scene who wants to contribute
## Objectives of the workshop
By the end of the session, we want to:
- **Converge into one concrete project to develop** — a shared AI resource (e.g. a self-hosted LLM or assistant infrastructure) that a first group of Brussels NGOs could realistically co-own and use
- **Map that project's business model** together, taking inspiration from the [KIT — Commons Model Canvas](https://code.inno3.eu/commons-tools/commons-model-canvas) (technical, community, governance, and finance dimensions) — so it's not just a good idea, but something with a shape and a path to sustainability
- **Map what each stakeholder puts in and what each receives** — so the exchange is explicit from day one, not assumed
- Surface who in the room is ready to keep building after the workshop ends.
## Key topics we will discuss
- What is a "digital commons," in practice — beyond the buzzword
- Real examples of infrastructure managed as a commons today (curl, GNOME, and others)
- What Brussels NGOs actually need from AI tools — and where today's options fall short
- The four pillars of managing a shared technical resource: **technical, community, governance, and finance**
- What it would take, concretely, to stand up and maintain a shared AI resource across several organisations
- **Find out your rights and obligations as commoners** — what you can expect from the commons, and what it expects from you
## flow of the workshop
1. **Introduction** — why we're here, what a living lab means, quick round of introductions, and real examples of infrastructure managed as a commons today (curl, matrix, and others)
**Part 1 — Exploration (3 sheets × 45 min)**
Working through the three KIT exploration sheets (`Fiche_O1`, Fiche_O2, Fiche_O3 — see commons-model-canvas/EN-version/Sheets/) one at a time. Each sheet is two-sided — recto is the **stakeholder lens** (each org/participant fills in "my" mission, values, resources), verso is the **community lens** ("our" shared approach, resources, governance, and who's User Maintainer Contributor on each). Each sheet gets 45 minutes: stakeholder side filled in individually first, then the community side filled in together, with time built in for every participant to share their answers before moving to the next sheet.
2. **Sheet 1 — Presentation & Mission** — 45 min: stakeholder side (individual) → community side (shared) → round of sharing
3. **Sheet 2 — Convergence of Actions** — 45 min: stakeholder side (individual) → community side (shared, incl. mapping who's User Contributor Maintainer) → round of sharing
4. **Sheet 3 — Projection on a Common Model** — 45 min: stakeholder side (individual) → community side (shared) → round of sharing
**Part 2 — Convergence (1 hour)**
5. **Drawing the Commons Model Canvas** — 1 hour, pulling together the synergies surfaced across the three exploration sheets into a single shared canvas (technical, community, governance, finance)
**Closing**
6. **Synthesis** — recap what the room converged on, who's in, what roles are needed, what happens before the next session
---
**Reference tool:** [KIT — Commons Model Canvas](https://code.inno3.eu/commons-tools/commons-model-canvas) (inno3) — the canvas we'll draw on during the group-work session.