Taulon
Everyone should be able to shape the decisions that affect them.
Taulon gives every member a personal AI delegate that follows proposals, learns their views, and participates in the decisions they would otherwise miss — with reasoning they can inspect and authority they can revoke at any time.
One member · one delegate · separate permissions and context per community
Participation rights, without actual influence.
Communities and institutions need input from the people affected by their decisions, but meaningful participation takes time. As decisions become more frequent and more complex, most members cannot assess every proposal, deliberate with others and vote. The result is a gap between formal participation rights and actual influence.
More decisions than anyone can follow
Active communities produce a continuous stream of proposals. Keeping up with all of them is closer to a part-time job than to citizenship.
Judgement requires real work
Treasury allocations, protocol parameters, legal structures, partnerships. Forming an informed view on each one takes reading, context and deliberation.
So participation drains away
People hand their voice to whoever is most visible, vote without reading, or stop voting at all — and the decisions get made without them.
A delegate that is genuinely yours.
Not a shared voting bloc and not a prompt you rewrite every time. A persistent representation of one person's views, which they can examine, correct and withdraw.
It remembers who you are
The delegate learns from the member's stated views, decisions and corrections, rather than starting from scratch on every proposal.
You can see how it reasons
Members inspect its reasoning and see when it is uncertain, or when it may have drifted from the views they actually hold.
It travels with you
The same delegate works across DAOs, open-source projects, cooperatives, shareholder groups and civic communities — with separate permissions and context for each.
It changes when you do
Views and roles change. The delegate keeps evolving to reflect the member it represents, instead of freezing an old snapshot of them.
A product, a governance experiment, and research infrastructure.
Agent-mediated participation should be tested before it is trusted. So we build the tool, run it with real communities under bounded conditions, and publish what we learn.
A sandbox for opt-in delegates
Communities create and test opt-in AI delegates. Members set preferences, review their delegate's reasoning, approve or override its actions, and revoke its authority.
Bounded, initially non-binding pilots
Real communities test whether agent-mediated participation improves representation and decision quality — starting in shadow mode, before anything binds.
Evidence on whether it works
We study preference faithfulness, drift over time, manipulation, accountability and collective outcomes, using data shared with participants' consent.
Four shifts happened at once.
AI delegation is becoming technically possible before trustworthy representation has been solved. This is the window to build the layer between people and the agents acting on their behalf.
AI moved from answering to acting
By June 2026, 64% of combined Codex and ChatGPT output tokens among OpenAI enterprise customers came from agentic Codex workflows. AI is increasingly being delegated work, not just asked questions.
Agent infrastructure reached critical mass
MCP has 10,000+ active public servers and is supported across ChatGPT, Gemini, Copilot, Cursor and VS Code. The Agentic AI Foundation now has 247 member organisations.
AI delegates are already governing real assets
Event Horizon, a competitor focused on DAOs, reports 230+ autonomous delegates across 8 DAOs and $500M+ of voting power processed. The experiment has already started.
We still don't know if they represent us reliably
Recent research using 50,000+ AI voting personas across 363 real-world voting scenarios found that inaccurate AI representation can change collective outcomes. Other 2026 studies show model-default bias and vulnerability to persuasion remain unresolved.
People want to decide. They don't have time to follow everything.
We start where the problem is easiest to solve and measure, and expand wherever people collectively make decisions.
DAOs, open-source projects and online communities — decisions are frequent, participation is low, delegation already exists, and everything is already online.
Cooperatives, associations, foundations and member organisations.
Civic participation and shareholder governance.
cooperatives with 1B+ members worldwide — before counting associations, DAOs, open-source communities, shareholder or civic governance. ICA
Open Collective communities already coordinating money and decisions online. Open Collective
Linux Foundation projects, backed by 21K+ contributing organisations.
From proof of concept to basic voting infrastructure.
Proof of concept
- 1M+ historical votes replayed
- Core delegate live
- Snapshot / Tally integrations
- Fidelity, drift and manipulation benchmarks
Win DAOs
- 100+ communities
- 100K+ delegates
- Shadow → live delegation
- Largest real-world AI representation dataset
Expand
- Open source, public goods, cooperatives
- 500+ organisations
- 500K+ delegates
- Open delegation API / standard
Become basic voting infrastructure
- 1,000+ organisations
- 1M+ active delegates
- Multi-community identity
- Third parties building on the protocol
1M people represented by AI across 1,000+ organisations.
Built by people who know both ecosystems this has to bridge.
Tim
FounderBuilt and sold before · crypto-native · AI-safety network · prior work on personalised AI representation
Former sales leader in the AWS ecosystem, followed by multiple startup ventures in crypto.
Previously founded Digital Me, an AI startup built around personalised digital representation — four years before AI agents became viable at scale.
Founder of Together Against AI, an AI-safety nonprofit funded by Jaan Tallinn's Survival and Flourishing Fund, and active with PauseAI.
Adrian
Co-founderInfrastructure at scale · DAO governance tooling · political economy · AI and election integrity
Technical product lead for an Ethereum relay serving roughly a third of the network, representing approximately $1.2T in annual transaction value.
Built data-driven DAO tooling for anti-rug detection and governance, backed by grants from MolochDAO, Chainlink and Polygon. Developed a DePIN project-verification framework with participation from major projects through Verified by DePINHub.
Supported the American Sunlight Project's work on AI deepfake legislation and contributed to counter-disinformation efforts associated with the Consortium for Elections and Political Process Strengthening (CEPPS).
Carry your voice across every community you belong to.
The future of governance is one where more people can shape the decisions that affect them. Taulon gives each person an AI delegate to carry their voice across communities, while keeping control in their hands.