selfdriven · the AI layer

Generative AI in service of human- and purpose-driven communities.

selfdriven AI combines the selfdriven Framework with the power of generative models — research, technology and practice for a world where AI acts under human governance and cryptographic accountability, not in place of it.

Explore

Four ways into the work.

The selfdriven AI layer is organised as a body of research, the technology that puts it into practice, the patterns for daily use, and the tools to build with.

Frameworks

Our published thinking.

Open frameworks for navigating a rapidly changing AI landscape — built to be used, argued with, and improved.

Our approach

AI that acts under human governance.

The selfdriven approach is the Human Conductor model: people define intent and hold accountability, while AI agents act under delegated, scoped, time-limited authority. The relationship is verifiable, not assumed.

The Human Conductor

Humans set intent, exercise judgment, and remain accountable. Agents handle volume, consistency and monitoring — never the things that should stay human.

Cryptographic accountability

Each agent operates under a KERI-delegated AID. Actions are anchored to an append-only key event log — independently verifiable, and durable enough to stand up to a regulator.

Purpose over scale

AI in the service of decentralised, human- and purpose-driven communities — measured by what it enables people to do, not by engagement or extraction.