2026-07-11 –, Creativity Lounge
Simocracy is a small experiment that lets you write a short brief for an AI agent and put it in a conversation alongside other agents and humans. This session is a discussion about what that does to a room. We will run a short live demo with sims debating a topic the room picks, then break out for small-group conversations about who writes these agents, whose perspectives they carry, and when you want them around. No background needed.
What this is
A discussion session about AI in group conversations. We bring a small agentic tool called Simocracy that lets people write a short brief for an AI agent, give it a point of view, and put it in a chat alongside humans. We run a short live demo so the room can see what AI deliberation actually looks like, then the rest is a guided conversation.
Who we are
A small team that has been building Simocracy for a few months. We have run two informal events with it. We are not pitching a product. We came to DWeb Camp because the people here ask better questions about this stuff than anyone else, and the questions we have are not technical ones.
What a sim is, in one paragraph
A sim is a short written document that says who an AI agent is: what it cares about, how it talks, what it knows. The same way you might brief a person before a meeting. Anyone can write one in about ten minutes. The agent then sits in a chat and responds in character. Two or three sims with different briefs end up disagreeing, the same way two or three people with different briefs would.
Why we think this is worth a session
AI agents are starting to show up in conversations where decisions get made: project meetings, funding committees, neighbourhood forums. Most of the time they are introduced from outside, by a vendor, and the people in the room do not get a say in how the agent behaves or what it cares about.
What we find interesting is when the room writes its own AI agents instead. The room decides which perspectives are missing and writes them in. The agents become more like puppets the room is honest about than like authorities pretending to be neutral. That changes who the AI in the room is for.
We do not know how this lands in practice. That is the part we want to discuss with you.
What happens in the 90 minutes
A short intro to what a sim is and how we got here, with no slides. Around ten minutes.
A live demo. We pick a low-stakes topic together, something like "should the camp have quiet hours after midnight" or whatever the room wants. The room helps write two or three sims to argue different sides. We run a short deliberation on the projector. Around twenty minutes.
Small-group discussion. Three prompts: who writes the agents, what perspectives are missing in your own communities, and when you want AI in the room and when you do not. Around twenty-five minutes in groups.
Report-backs and an open conversation about what came up. Around thirty minutes.
A short close with a link to the tool if anyone wants to try it. Around five minutes.
What we want out of it
Sharper questions. We are especially curious where you think this is dangerous, what the obvious failure modes are, and what kind of community might actually find it useful.
What we need
A room for around thirty people, a projector, and one big screen. Power for a couple of laptops. Network is helpful but not required.
David Dao is Chief Scientist at GainForest.Earth and Econ & Gov Lead at Protocol Labs
A pioneer in data valuation for machine learning, David holds a PhD in AI Systems from ETH Zurich. His work, which sits at the frontier of AI and decentralized systems, has been globally recognized by the World Economic Forum, XPRIZE, and Ethereum.
- Broad Listening: What changes when AI sits in on the conversation?
- Designing Funding Mechanisms on AT Protocol with Hypercerts
- GainForest: How we can ensure equitable and regenerative intelligence for nature conservation?
- How to Sustain Open Source Software in the Era of AI? A Community Solution of Prosperous Software Movement
Holke Brammer is a political economist working to strengthen collective action through open technology. As founder and director of Hypercerts Foundation, he leads the development of open infrastructure for impact funding and public goods resource allocation. He was previously a research scientist for public goods funding at Protocol Labs, worked at Yunus Social Business and Boston Consulting Group, and led Open Social Innovation efforts in Germany with ProjectTogether. He holds a Ph.D. in Political Economics from Stanford Graduate School of Business.