Onecub
FR Let’s discuss your projectContact

Trust is built.
Governance keeps it alive.

To share data and coordinate AI agents, partners need to know what they commit to, and how their commitments will be followed up.

Governing a data space also means deciding how AI accesses and uses data.

Data access, purposes, agent mandates, human approval: these choices belong to the group’s governance, just like its roles and its decisions.

From individual choices to shared rules.

  1. People

    Personal choices

    Everyone decides what they share, with whom and for how long.

  2. Organisations

    Internal usage rules

    Each organisation sets what its teams and AI tools may do with data.

  3. Ecosystems

    Shared rules

    The group agrees on the rules that apply to all its participants.

We work at all three levels, so that individual choices and the group’s rules stay consistent.

From objective to evidence.

A rule is only worth something if someone applies it and can show it was followed. We connect each of the group’s objectives to these six links.

Human control
A person checks or approves.
Contractual control
A written commitment, and its remedies.
Technical control
A mechanism that applies or checks the rule.

Writing a policy does not guarantee it will be enforced automatically: each rule needs an owner and a means of control.

  1. ObjectiveWhat the group wants to make possible.
  2. RuleWhat is allowed, and under what conditions.
  3. ResponsibilityWho applies the rule and who answers for it.
  4. ControlA human review, a contractual commitment or a technical mechanism.
  5. EvidenceThe record that shows the rule was followed.
  6. ReviewThe rule evolves with use and experience.

An example: a request to access data.

  1. Request.A participant requests data for a specific use.
  2. Conditions checked.The rulebook’s conditions are checked: the requester’s role, purpose, duration.
  3. Access within a defined scope.Access is granted for this purpose and this duration, and no further.
  4. Record.The decision and the use are recorded, so they can be accounted for.

A teaching example: it illustrates a principle, not an available product.

Governing the group, not just access.

An ecosystem lives on its shared decisions. Its governance must be able to welcome new members, settle disagreements and evolve its rules without renegotiating all its foundations.

  • Decisions. Who decides, by what majority rules, and in which bodies.
  • Representation. How each category of participants makes its voice heard.
  • Contributions. What each brings, and what each gets back.
  • Disagreements. How they are handled, then settled.
  • Changing the rules. Who can propose a change, and how it is adopted.

Changing a rule.

  1. Propose.A participant or a body proposes a change.
  2. Discuss.The participants concerned give their views.
  3. Approve.The competent body adopts, amends or rejects.
  4. Version.The new version is dated, and the previous one remains available.
  5. Inform.The participants concerned are told about the changes.

An example of a process, to adapt to each group.

Reusing rules across ecosystems.

Start from shared rules, adapt them to your context and keep a record of your decisions.

Local rules stay compatible with the obligations and common commitments that apply. This is what lets autonomous ecosystems cooperate.

A three-step method.

Frame

Map stakeholders, use cases, decisions and risks. Identify existing rules and the points to settle.

  • Role mapping
  • Diagnosis
  • Priorities

Formalise

Co-design the group’s rules, onboarding journeys and responsibilities. Keep the rulebook, contracts and permissions consistent.

  • Governance model
  • Rulebook
  • Participant journeys

Put to the test

Test the framework on real scenarios: a newcomer, a change of purpose, an incident, an agent’s action. Plan controls, evidence and rule reviews.

  • Test scenarios
  • Control plan
  • Review cycle

Does your group need a rulebook, or need to evolve one?

Let’s discuss governance

AI acts.
Who gives it a mandate?

With AI agents, governance extends from data to actions. An access authorisation is not enough to define what an agent may decide, on whose behalf it acts, or when a person must take back control.

We address these questions from the design stage: scope of delegation, limits of action, supervision and accountability. A framework to test with your teams before expanding the uses.

Agentic AI for data spaces

Open to cooperation.
Fitted to your ecosystem.

We build on European data space work and open standards. The framework reflects your sector, your partners and your business model, without imposing a single technology.

We are also developing a tool to help groups keep their governance alive.

Coming soon: steer your ecosystem’s governance

Rules that hold, for cooperation that lasts.

Let’s talk about your group, its decisions and its commitments.

Let’s discuss governance

Or write directly to olivier@onecub.fr