CivicSuiteWindows Local

One installer · six modules · the city's own computer

Everyday city hall work, on a computer the city actually owns.

CivicSuite Windows Local is an offline operations suite for small cities. One Windows installer sets up the whole city-core suite — six connected modules, their shared database, and a local AI model — so a clerk can do real work with no cloud account, no telemetry, and no per-seat pricing. Current release: v1.0.2.

Download CivicSuite Windows Local Read the manual v1.0.2 · GA candidate · public beta
  • Local firstNo cloud requirement, no telemetry, no per-seat pricing. Records stay on the city's machine.
  • One installer, whole suiteA single MSI installs all six city-core modules, the database, services, and the local AI model together. No Docker, WSL, or terminal — and no per-module shopping.
  • Honest about statusA GA-candidate beta. The one remaining gate to General Availability is a code-signing certificate.

Status · 2026-07-02

Where the project is today

Current build: CivicSuite Windows Local v1.0.2 GA candidate · public beta — predecessor v1.0.1.

What "GA candidate, public beta" means. The build is feature-complete for the city-core profile and was validated end to end on a clean Windows machine. The MSI is Authenticode code-signed via Azure Trusted Signing. This is not a public-use, city-ready, procurement, production, or full-suite promotion claim. Beyond city-core, public shipping, product-ready, and v1.0 maturity claims are frozen until each module repo re-earns them through the release-recovery gates.

The city-core profile

Six modules, one shared platform

You don't install modules one at a time — the installer ships all six as one suite. First-run setup defaults to the full City Core profile; CivicCore is the shared platform every module depends on and is always installed. The modules share one identity, one audit trail, and one search. Each module name below links to its source repository on GitHub for the deeper dive.

v1.2.0

Shared platform: identity, audit, retention, the local task queue/worker, and document ingestion. Every module depends on it; it depends on none.

Public-records / FOIA request intake, search, and AI-assisted response drafting.

Meetings, agendas, packets, minutes, and votes.

v1.0.8

Municipal code: searchable ordinances with AI-assisted guidance.

Public-notice creation and publishing workflow.

Accessibility workflows and records-ready export: plain-language rewrites, translations, and WCAG review support.

CivicZone, CivicPlan, CivicPermit, and CivicInspect remain queued Tier 2 modules and are not part of city-core. The wider catalog is foundation-tier (schemas and specs), not city-ready product.

What's new in v1.0.2

The headline: the CivicAccess Accessibility tab

Every release updates the whole suite. v1.0.2's headline is the on-screen Accessibility workflow tab: alongside deterministic checklist tools (form plans, publishing workflow, ADA Title II, tagged-PDF expectations), it introduces three features that run on the bundled local Gemma model:

Plain-Language Rewrite

Drafts a plain-language version of text a clerk enters.

Multilingual Variant

Drafts a translation of the entered text into any target language.

Accessibility Review

Runs deterministic WCAG rule checks and stores a short model-drafted remediation analysis alongside them.

How the AI is scoped — read this. Each AI feature is dual-path: when the local model isn't ready, it falls back to a deterministic result clearly labeled "AI engine not ready." Every AI output is labeled a draft requiring human review. Translations must go to a qualified human translator before public use. A saved Accessibility Review is advisory clerk support, not a certified accessibility audit — the records-bearing status derives only from the deterministic rule checks; the AI never adds, removes, or reclassifies a finding.

Two ways to read this

For the people who run it

For clerks & decision-makers

You do not need an IT department. CivicSuite installs the way any Windows program does, then runs on that computer. No website to log into, no monthly subscription, and nothing about your city's records leaves the machine.

  • Your data stays yours. Records, minutes, notices, and requests live on your own computer — no cloud service to go dark, change price, or be subpoenaed out from under you.
  • No recurring bill. No per-seat licensing and no per-user fees; it runs on hardware you already own.
  • Works offline. The AI runs from a model bundled inside the app — not a paid API — so it keeps working with no internet.
  • AI that stays in its lane. It drafts and suggests; a person always reviews and decides.

Recommended hardware: 32 GB of RAM (16 GB is a workable minimum), about 15 GB free disk, and 64-bit Windows 10 or 11 with WebView2. Most office desktops bought in the last few years qualify.

For IT & technical evaluators

A single Tauri/WebView2 desktop app that bundles its full runtime — no external service dependency, no container runtime, no developer tooling on the target machine. Everything runs as local processes supervised by the shell.

  • Database: portable PostgreSQL 17 with pgvector — fully bundled; first-run setup completes on a factory-fresh Windows PC with no prerequisites.
  • Services: embedded CPython city services with a PostgreSQL-backed local task queue and worker (the Windows-local profile, replacing the server-profile Redis/Celery stack).
  • Local AI: Ollama serving a pinned local Gemma model, registered through an explicit download / checksum / runtime step on first run — on-device inference, no API.
  • Trust path: verify the published MSI SHA-256 and per-module source_commit pins (PROVENANCE, signing policy). The MSI is Authenticode-signed via Azure Trusted Signing. A new certificate has no reputation yet, so SmartScreen may still prompt on first run — click More info, confirm it shows a verified publisher (not "Unknown Publisher"), then Run anyway.

Architecture

Everything runs on one machine

CivicSuite Windows Local topology All components run locally on one Windows machine: a Tauri/WebView2 desktop shell drives embedded Python services, which talk to a bundled PostgreSQL 17 with pgvector database and to Ollama serving a local Gemma model. Nothing leaves the machine. ONE WINDOWS 10 / 11 MACHINE · NO CLOUD · NO TELEMETRY Tauri / WebView2 desktop shell & UI — six modules Embedded Python services bundled CPython · local task queue PostgreSQL 17 + pgvector portable · fully self-contained records, minutes, notices, audit, vectors Ollama + local Gemma pinned model · downloaded & SHA-verified on-device inference — no API local IPC SQL HTTP · localhost
The desktop shell drives embedded Python services; those talk to a bundled PostgreSQL 17 + pgvector store and to a local Ollama instance serving a pinned Gemma model. No traffic leaves the machine.
Tauri / WebView2 desktop shell PostgreSQL 17 + pgvector Embedded CPython services Ollama + Gemma on-device Local only · no cloud

Where to go next

Documentation & trust path