Your ballot. Your representatives. Your data. Understood in minutes.

We built an open-source platform that turns scattered government data into civic tools communities own and control. AI-powered analysis runs entirely on your infrastructure — no data leaves, no third parties involved.

The problem is access, not information

A California ballot proposition is 47 pages of legal text. A city council meeting agenda references a dozen municipal codes. Campaign finance data is spread across multiple state databases in formats designed for compliance, not comprehension.

The information is public. But the platforms that promise to make it accessible track what legislation you research, profile your political interests, and route your queries through third-party AI services. Your civic engagement becomes someone else's data product.

We built the alternative: tools that serve citizens, not the other way around.

Community-owned

Infrastructure you own, not a platform you rent. Every line of code is open source under AGPL-3.0.

Privacy by design

All AI runs locally on your hardware. No data leaves your infrastructure. Location data is geo-fuzzed. Logs mask PII automatically.

Built to last

Four microservices, 16 pluggable provider packages, 1,000+ automated tests, and production deployment configs.

Accessible to all

WCAG 2.2 AA compliant. Works offline as a progressive web app. English and Spanish.

Why this works now

Three technical shifts made community-owned civic intelligence possible.

01

LLMs parse legal text

Open-source language models can now read legislation, extract fiscal impacts, and generate plain-language summaries that rival expert analysis. We run Ollama locally with models like Qwen and Mistral — no API calls, no per-query costs.

02

Vector search enables civic RAG

Upload a 200-page budget document and ask questions in plain English. Embeddings generated locally with Transformers.js, stored in pgvector, and retrieved with semantic search — the entire pipeline runs on your hardware.

03

AI scraping self-heals

Government websites change layouts constantly. Our schema-on-read pipeline uses AI structural analysis to generate extraction rules automatically. When a site changes, the pipeline adapts — no developer intervention required.

How it works

01

Choose your jurisdiction

Pick your state, county, or city. The platform loads your local civic data automatically through region plugins — JSON configurations that tell the AI what to collect.

02

Explore what matters

Scan petitions with your camera, read propositions in plain language, follow campaign money, look up your representatives by address.

03

Take action

Email your representatives directly from the platform, attend public meetings, share analysis with your community. Everything works offline.

Who it's for

Built for the people and organizations that make democracy work.

Community organizations

A neighborhood association preparing voters for a local bond measure. A civic group tracking campaign spending on a ballot initiative. Direct access to the information your community needs.

Get Involved →

Developers

Add your jurisdiction in a weekend with JSON configs — no scraper code required. The AI-powered pipeline figures out how to extract data from government websites automatically.

Contribute →

Local governments

Make council meeting records searchable. Give residents plain-language access to municipal codes and ballot measures. Increase transparency without building custom software.

Learn More →

AGPL-3.0

License

California, 2026

First pilot

4 microservices

Architecture

100% local

AI approach

Ready to build with us?

Bring citizen-owned civic technology to your community. Deploy the full stack on your own infrastructure, or contribute to the platform that a growing network of communities depends on.