VoiceTrace

Community voices you can check.

Every year, cities, school districts, and government agencies ask communities for input. Residents fill out surveys, attend meetings, and share what matters to them. Then a summary appears and a decision gets made. But the path between the two is opaque. Did your voice actually shape the outcome? There is currently no way to know.

VoiceTrace is a provenance architecture for stakeholder engagement that makes that path traceable. Rather than asking communities to accept that their input was heard, it gives them the infrastructure to verify it.

The Problem

The synthesis step in civic engagement, where raw community feedback becomes actionable insight, is a black box for everyone involved. Civic leaders work under pressure to analyze hundreds of responses with no systematic way to justify their interpretations. Community members have no way to check whether their specific perspective shaped any outcome. As AI increasingly mediates this synthesis, the question of whether you are engaging with real voices or AI-generated artifacts has become more urgent. Transparency is not enough. What large-scale engagement needs is verifiability. This challenge is not unique to civic contexts: any organization that collects input at scale, from a city running a public hearing to a company running employee feedback processes, faces the same broken feedback loop.

What We Are Building

VoiceTrace is in active development and comprises three integrated components.

Provenance architecture: A structured data model that links every synthesis output back to the specific voices supporting it, and traces those voices back to the documented purpose that shaped the questions asked in the first place. The pipeline processes community input across four layers: project context captured at elicitation, verbatim community voices, structured items and argument chains extracted from those voices, and synthesis outputs. The provenance chain runs in reverse, grounding every synthesis output in traceable citations back to specific voices and the engagement context that produced them.

Sensemaking interface for civic leaders and facilitators: A structured workspace where planners analyze community input and generate synthesis outputs, each grounded in traceable citations to real voices. Every insight is auditable and defensible.

Community platform: A layered public interface designed for different levels of engagement. Casual visitors get accessible narrative summaries. More engaged readers can verify that voices are real and traceable to actual engagement events. Deep divers can follow the full chain from a synthesis insight back to the specific community member who shaped it.

Why It Matters

VoiceTrace is the first system to trace stakeholder voices end-to-end across the full engagement lifecycle, from elicitation through synthesis to outcome, auditable by both the leaders who run the process and the people who participated. We are currently focused on civic contexts, where the stakes of a broken feedback loop are highest: school redistricting, neighborhood planning, and public resource allocation are decisions that shape people's lives directly. But the infrastructure we are building extends to any domain where voices are collected at scale and decisions follow, including employee feedback processes in large organizations where the same verification gap exists.

The goal is engagement infrastructure where both groups can trust but verify: where participants can follow their voice from the moment they submitted it to the decision it helped shape, and where facilitators can demonstrate, not just claim, that synthesis was systematic and grounded in real input.