Open Source Civic Intelligence
Your government,
in your language,
in plain language.
Government documents go in. Plain-language civic briefs come out. In the languages your community actually speaks.
Try it now →City Budget FY2026
Resolution No. 2026-0142
Zoning Amendment 47-B
Civic Brief
Budget
Property Tax Up 8.2%
$4.1M increase allocated to road repair and school facilities.
Zoning
Elm St Rezoned Mixed-Use
Blocks 400-600. Approved 4-1. Comment deadline: April 5.
Policy
New Farm Subsidy Scheme
Farms under 5ha eligible. Apply by May 1.
3,500+
newspapers closed in 20 years
213
US counties, zero local news
50M
Americans with limited civic access
4B+
citizens in non-dominant languages
Five scenarios. One platform.
Click each tab to see both sides: what goes in, and what comes out.
City Budget FY2026-27
DEPARTMENT OF FINANCE
PROPOSED BUDGET FISCAL YEAR 2026-27
General Fund Revenues: $142,800,000
General Fund Expenditures: $148,200,000
Sec 4.2 Property Tax Levy:
Current rate: 1.12 per $100 assessed
Proposed rate: 1.21 per $100 assessed
(8.2% increase over FY2025-26)
Allocation Changes:
Public Works: +$4,100,000 (road repair)
Education: +$2,300,000 (facilities)
Public Safety: -$890,000 (fleet defer)
PROPOSED BUDGET FISCAL YEAR 2026-27
General Fund Revenues: $142,800,000
General Fund Expenditures: $148,200,000
Sec 4.2 Property Tax Levy:
Current rate: 1.12 per $100 assessed
Proposed rate: 1.21 per $100 assessed
(8.2% increase over FY2025-26)
Allocation Changes:
Public Works: +$4,100,000 (road repair)
Education: +$2,300,000 (facilities)
Public Safety: -$890,000 (fleet defer)
AI
Civic Brief
Your Property Tax
Going up 8.2%. A $300K home pays ~$270 more/year.
Where It Goes
$4.1M to road repair. $2.3M to school buildings.
What Got Cut
Police fleet replacement delayed. Saves $890K.
FY24 FY25 FY26
Comment By
April 12. Budget hearing: April 18, 7pm City Hall.
How it works
From bureaucratic PDF to civic clarity in minutes.
Ingest
Monitors government feeds for new budgets, legislation, meeting minutes, and public notices.
Interpret
AI with civic-context prompting: what changed, who is affected, where the money goes.
Verify
LLM-as-Judge scores accuracy. Low-confidence outputs routed to community reviewers.
Translate
Simultaneous output in English, Spanish, Hindi, Hausa, and any community language.
Deliver
Mobile portal, WhatsApp, newsletters, embeddable widgets. Where people already are.
AI that earns civic trust
Getting civic information wrong is not a quality issue. It is a democratic harm. Four layers of verification.
1
Source Grounding
Every summary generated solely from the source document. No general AI knowledge for civic claims. Auditable.
2
LLM-as-Judge
Separate AI model scores factuality against source. Below threshold? Flagged for human review before publication.
3
Community Review
Civil society partners and local experts verify flagged summaries. Every correction improves future output.
4
User Feedback
Citizens flag errors directly. Patterns of error inform systemic improvements to civic prompting.
“Transparency is not a luxury. It is the minimum requirement for democratic participation.”
Jatin Patel, Founder · 20+ years · Group PM, Microsoft Teams · Built PeopleBuilt.ai solo
Open by design. Day one.
Every line of code, every civic summary, every document processed. Public, auditable, deployable by any community, anywhere.
MIT License
Next.js 16
TypeScript
Supabase
Claude API
Vercel
Questions? civicbriefapp@gmail.com