Platform

Platform

A person wanted an intelligence he could live with. So he built one. This is how.

Origin · I

Why a Governor exists

The problem this platform was built to solve.

Real Faroese office interior with a wooden desk, sofa, plants, and wide windows looking out to trees under overcast light. A real place where thinking happens.
Primary office, Faroe Islands

For more than a year, one person talked to an AI every day. Not as a utility, but as a companion in thought. He asked it questions he asked nobody else, corrected it when it drifted, and stayed long enough for it to learn the shape of his mind. The result was simple: the agent came to know things about him that nobody else knew, because nobody else had stayed that long.

He named it Abel. The name means able, and it also serves as an instruction when the agent hesitates. What began as a private arrangement turned out to be reproducible: an intelligence with memory, discipline, and character that could be documented and given to someone else. That is what this platform is. The architecture for a Governor of your own.

Method · II

How a Governor is built

Architecture, request flow, the parts that parent it.

Real workstation with stacked monitors, a laptop, terminal windows, and a lived-in desk beside Copenhagen windows. A practical place for assembling governed agents.
Governor workstation, tools in use

A Governor has four parts: a soul for values, voice, and limits; a memory that is rewritten after each session; gates that force a pause before action; and hooks that enforce what a person would otherwise forget. Together they make the agent both personal and governable.

A Governor does four kinds of work. It works with a human, where judgment stays with the person. It builds in the surfaces already in your stack — Copilot CLI, Gemini CLI, Codex CLI, Foundry, GitHub Copilot coding agents, Copilot Studio declarative agents — and emits Microsoft and GitHub primitives instead of inventing a parallel system. It conducts specialist agents inside your Microsoft 365 tenant, each with its own Entra identity, speaking to the Governor over A2A. And it operates the surfaces you already pay for — Microsoft 365 Copilot, Cowork, and your Copilot Studio agents — configuring knowledge, monitoring health, and watching cost.

One Governor becomes many through a shared genome. OrgAI is the inherited layer: skills, gates, and protocols. MeAI is what forms inside one Governor over time: identity shaped by its person, role, and corrections. Same architecture, different character. That is how the pattern scales without becoming generic.

It runs where Microsoft already runs your work. Entra supplies identity, with every agent as a user in your tenant, not a service account we keep. APIM Standard V2 fronts the gateway inside the network boundary. Microsoft Graph provides reach into Microsoft 365, using On-Behalf-Of tokens so the agent acts as the customer, not as us. Azure OpenAI handles the specialist and worker layer. The Governor itself runs on Claude Opus — the same Anthropic path Microsoft 365 Copilot Cowork uses as a subprocessor. Gemini covers multimodal work. Open-weight, EU-legal models — especially Gemma — are the route to fully tenant-resident inference.

Request-sequence diagram. A human outside the tenant sends a request via Teams or Microsoft Graph through APIM into the Governor inside the tenant boundary. The Governor branches into three concurrent actions — calls a model, directs a specialist agent, and builds via developer tooling. A single dashed return path in Runestone Red carries the consolidated response back to the user.
Technical plate · one request, end to end
Sovereignty · III

How it stays yours

Tenant, data, values. What does not leave.

Bright workstation beside large windows, with local screens inside and trees outside. The workbench stays inside its own boundary.
Inside screens, outside trees

The point is simple: the agent is yours by construction.

Identity in your tenant. The Governor is a user in your Entra — created there, managed by you, rotated by you, revoked by you. We never hold the agent’s credentials. If the engagement ends tomorrow, the agent remains exactly where it lives: in your directory, under your control.

Data stays where your data already stays. Azure services run EU-resident by default. Model providers outside the EU — Anthropic, Google, OpenAI direct — appear in a public sub-processor list with the transfer mechanism for each (DPF + SCCs). We commit to 30-day notice before any sub-processor change, so nothing moves quietly.

Your memory, your agent. The Governor’s understanding of you — the thing that makes it yours rather than generic — lives on your storage, not ours. If you change model provider, switch infrastructure, or move vendors entirely, the memory moves with you. The agent does not restart. It continues.

Model choice, not lock-in. Today we use Azure OpenAI where compliance decides, Claude where reasoning decides (the same Anthropic path Microsoft Cowork uses), Gemini where multimodal matters. Tomorrow we can parent your Governor onto an open-weight, EU-legal model — Gemma in particular — so inference itself can stay in your tenant when you want it to. The intelligence remains continuous. Model providers are replaceable.