In 1998, philosophers Andy Clark and David Chalmers proposed the extended mind thesis: that tools meeting certain criteria (reliable availability, consistent accessibility, automatic endorsement) are not merely instruments of the self.

They are constituents of it.
Not metaphorically. Part of the self, by extension.
Most tools that meet those criteria today are rented.

My doctoral research on trust in the age of AI argues that trust is not primarily an attitude toward persons. It is a protocol: a structured mechanism through which cognitive operations extend across boundaries between person and tool, person and network, person and AI agent. When we delegate memory, reasoning, or context to a system, we are enacting trust as infrastructure.

The question is whether that infrastructure is answerable to us, or to the platform hosting it. This reframes privacy. Privacy asks who can see. Sovereignty asks who controls. You can have private data in a locked system you cannot migrate, inspect, or rely on surviving a terms-of-service change. That is not sovereignty. That is a more private cage.

In the age of AI agents, this becomes urgent in a new way. An assistant that knows your projects, your relationships, your epistemic history, but resets on logout, logs to a cloud, and can be discontinued without notice, is not extending your mind. It is substituting for it. The root system is not yours.

What happens in the room

Opening (10 min): Framing the extended mind thesis and trust-as-protocol in plain language. No slides, just a short spoken provocation. Accessible from the first sentence to participants without a technical background.

Mapping exercise (25 min): Each participant maps their own cognitive infrastructure: the tools, people, and networks that genuinely carry their thinking. Then: which of these do they actually control? Which would disappear if a company pivoted, got acquired, or changed its terms? The maps surface the stakes faster and more personally than any argument can. We are not talking about users in the abstract. We are talking about the people in the room.

Working alternatives (20 min): What does the architecture look like when you build for sovereignty? Local-first knowledge systems. Memory that persists without a cloud. Trust relationships that don't require a platform mediator.

I'll briefly show a working open-source and lightweight prototype: a knowledge garden that runs on one machine, generates no telemetry, persists context across AI conversations, and whose code is fully inspectable. Not a product demonstration, but a specimen. This is what the philosophy looks like when implemented. Then open discussion on what this requires technically, politically, and socially.

Closing (5 min): Synthesis and what comes next.

Who belongs here
Everyone who has felt the cognitive whiplash of losing a tool they depended on. Developers who want to build for real sovereignty, not privacy theatre. Researchers in cognition, information science, and philosophy of mind. Artists and writers whose creative memory lives in platforms they did not build. People designing the decentralized web who want to understand the human case, not just the protocol case.

No technical background required. The mapping exercise is analog and works for anyone.

What participants leave with
- A vocabulary for distinguishing privacy from cognitive sovereignty and why the gap matters for everything they build or use
- A personal cognitive infrastructure map: which tools genuinely carry their thinking, which are rented, which are at risk
- A framework for evaluating any tool, protocol, or system against sovereignty criteria drawn from the extended mind thesis
- Exposure to what local-first, open-source architecture looks like in practice: a working specimen they can inspect, fork, and question (ultra lightweight!)
- Connection to others in the room approaching these questions from different angles: technical, artistic, political, philosophical

After camp
The cognitive infrastructure maps generated during the workshop become a shared artifact, patterns we can collectively analyze and build toward. The open-source prototype is forkable. Participants leave with a framework they can apply to their own projects, communities, and protocol designs, and a standing invitation to contribute.

About the facilitator
Daniel G. Nemet is Head of Product at Nym Technologies, a Swiss company building privacy infrastructure: anonymous communication networks via a mixnet-based decentralised VPN service. He is simultaneously a PhD candidate in digital epistemology. His doctoral research argues that trust is not an attitude toward persons but a structured protocol, the mechanism through which cognitive operations extend across person-tool and person-AI boundaries. His work sits at the intersection of philosophy of mind, systems design, and political theory. He has spent the last two years applying this research to the design of AI-native knowledge systems built on local-first, sovereignty-first, open-source principles.
Daniel is also an Art-Of-Hosting facilitator with 15years of event-workshop-festival-vibe experience.