Government makes the claim.
We check the ground.
CitizenFirst.ai builds AI tools that cross-reference public spending records against satellite imagery, street-level evidence, and citizen reports — so "sanctioned" and "built" stop being different things.
Citizens fund public infrastructure. They rarely get to see what it bought.
Every year, road schemes, school grants, and welfare budgets are sanctioned on paper. The money moves. The paperwork closes. But whether the road got tarred, the school got its roof, or the ration reached the family — that's left to trust.
CitizenFirst.ai starts from a simple premise: a government record is a claim, not a fact. Claims should be checkable — by anyone, not just auditors with insider access. We build tools that make checking possible at the scale a country actually needs, using data that's already public but never cross-referenced.
We're not building a watchdog that shouts. We're building the instrument that lets citizens, journalists, and honest officials point at a map and know, in seconds, whether a claim holds up.
This is day one. There's one flagship concept below, a short roadmap, and an open invitation to help shape what comes next.
The Road Ledger — India's rural roads, sanctioned vs. built
India sanctions thousands of crores a year for roads under schemes like PMGSY and state PWD budgets. The Road Ledger cross-references two layers that have never been paired at scale: sanction and expenditure data (from PMGSY's OMMAS portal, GeM procurement, RTI responses) against satellite and street-level imagery that can distinguish a tarred road from a mud track, and flag missing railings or lighting.
Concept stage — not yet live↑ Illustrative mock-up of the intended output — no live data yet. Real entries will link back to source documents.
Three layers, cross-referenced automatically
Pull the paper trail
Scrape and structure public sanction and expenditure data by GPS coordinate — OMMAS, state PWD portals, GeM, and RTI responses.
Read the ground truth
Classify road surface from satellite bands (tar vs. mud has a distinct spectral signature) and check street-level imagery for potholes, railings, and lighting.
Show the gap
Overlay both layers on a map. Green where the claim holds. Red where money moved but nothing visible changed. Every entry links to its source.
Roads first. The pattern generalizes.
Once the verification pattern works for roads, the same architecture — public record vs. ground evidence — applies almost anywhere government makes a checkable claim.
Stalled Projects Tracker
A public dashboard for infrastructure projects stuck between ministries — sanctioned, funded, and going nowhere.
School & Welfare Ledger
Extending the same claim-vs-evidence model to education grants and welfare scheme delivery.
Open Verification API
Letting journalists and researchers query the ledger programmatically instead of scraping it themselves.
This is one person's idea right now. It works better as many people's.
We need people who can wrangle satellite imagery, people who know India's public data portals better than we do, people who've done civic-tech in other countries and know what breaks at scale, and people who'll just tell us honestly if this is worth building.
No pitch deck, no funding round yet — just a working idea and a domain. If that's interesting to you, say hello.