What does your
community
actually want?
Three layers of intelligence
From individual perspective to collective understanding in three transformations.
THE INTERVIEW
High-Fidelity Signal Capture
Users converse with an AI mediator instructed to use Motivational Interviewing and Nonviolent Communication techniques. No surveys. No polls. A real conversation that draws out underlying values, not just surface positions. The transcript stays in your browser.
“A "No" on a bike lane becomes: "Supports cycling infrastructure conditionally. Requires: parking mitigation. Core value: economic vitality."”
THE SYNTHESIS
Consensus Crystallization
Each interview becomes a structured summary of values, needs and conditions. Participants are grouped by value profile, and shared concerns are grouped by underlying need — not by for/against. Support is counted from participants, never estimated by the model.
“"Defund the police" and "Back the blue" both contain people who want safe neighborhoods with accountable public servants. The Civic Filament finds that signal.”
THE DOCUMENT
Living Requirement Document
The output is a versioned document, regenerated as perspectives arrive: Strong Consensus (70%+ and at least 5 people), Conditional Consensus, Genuine Divergence, and Confidence Gaps. Nothing is published until at least 10 people have taken part.
“Like version control for community intent — every synthesis run creates a new numbered version, older versions stay viewable, and each version publishes a Merkle root of exactly the perspectives it used.”
The conversation, not the survey
A three-turn sample from the bike-lane demo issue. Every line below is pre-written — this is a simulation of the real interview, not a live LLM call. When you’re ready to actually talk to the mediator, head to /participate.
Scripted three-turn interview preview: Main Street bike lane demo issue
Mediator: The city has proposed a protected bike lane on Main Street, replacing 12 curbside parking spaces. Before we talk about whether you'd support it — what matters to you about how Main Street works day to day?
Participant: Honestly? I run a small shop on Main and half my regulars park right out front. If they can't stop for five minutes on their way home, they drift to the strip mall. It's not about bikes versus cars — it's about my rent this year.
Extracted signals: VALUE — Economic security; NEED — Predictable customer access.
Mediator: Thank you — that's helpful. If the parking loss were addressed — a loading zone kept nearby, or side-street spots reconfigured — would your read on the bike lane itself change?
Participant: Maybe. I'd still want to see what the actual safety data says. I'm not against bikes — I've got two kids who ride to school on that stretch.
Extracted signals: VALUE — Child safety; CONDITION — Parking preserved.
Mediator: That's useful. Is there a version of this the city could propose that would make you a clear yes?
Participant: Give us a real loading zone, publish the crash data for that block, and phase it in over a year so businesses can adjust. I could get behind that.
Extracted signals: POSITION — Conditional support; CONDITION — Phased rollout; CONDITION — Transparent data.
In a real deployment, this response would join a Conditional-Consensus cluster in the Living Requirement Document. See the Methodology page for how clustering works.
Values and needs (not positions) are what get clustered. That’s how “defund the police” and “back the blue” can end up in the same cluster around accountable neighborhood safety.
How consensus crystallizes
As each perspective enters the system, the consensus map restructures. Clusters of shared values emerge, connect, and strengthen — revealing the architecture of collective intent that was always there, just invisible.
The visualization to the right is a scripted illustration built from the bike-lane demo issue — not live participant data.
One protocol, infinite applications
City Planning
Replace contentious town halls with structured dialogue. Surface what residents actually agree on before the shouting starts.
Decentralized Orgs
Move beyond "vote on everything" governance. Understand member values before writing proposals. Ship what the community actually wants.
Continuous Listening
Replace annual engagement surveys with structured, ongoing sentiment synthesis. Understand what your workforce needs — in their own words, not on a 1–5 scale.
Public Opinion
Media organizations and researchers replace binary polls with consensus maps. Show the public the nuance that actually exists in their own opinions.
Scale with your community
For grassroots organizers, small communities, and researchers.
- ◇Free to participate; community issues launched with us on request
- ◇Full Living Requirement Document
- ◇Public dashboard + LRD version history
- ◇Community-supported (feedback form)
For city governments. Launch community consensus on any issue, with no per-issue fee.
- ◇Unlimited issues; up to 3,000 interviews per month included (more on request)
- ◇Full LRD with provenance verification
- ◇Customer portal: request issues, track usage, download council reports
- ◇Pilot onboarding
- ◇Dedicated support
- ◇Written Data Handling Addendum per jurisdiction
For state/regional bodies and large organizations. Custom scope.
- ◇Everything in Municipal
- ◇SSO, EU hosting and multi-jurisdiction deployments: on the roadmap, scoped per contract
- ◇No on-premise deployment
- ◇Dedicated account management
Price framing: one consultant-led community engagement study for a single issue typically runs $20,000–$50,000. The Municipal tier runs unlimited issues under that number.
The question “what do people want?”
deserves a real answer.
Start with one issue. One community. See the consensus that was always there, waiting to be found.
FREE TO PARTICIPATE // COMMUNITY ISSUES LAUNCHED WITH US ON REQUEST
A low-frequency newsletter is planned for changelog notes and pilot-slot announcements. It hasn’t launched yet — leave your email and we’ll only use it to notify you when it does.