The AI operating system
for cities.
One platform. One city model. Unlimited decisions.
Cities buy infrastructure blind.
Every utility holds its own data and none of them connect. So the question that actually matters at budget time — what breaks next, and what is the cheapest thing that stops it — has no owner and no answer.
Every building. Every road. Every resident.
We tell you which numbers are real.
| Layer | Source | Status |
|---|---|---|
| Buildings & roads | OpenStreetMap, extracted per district | Measured |
| District population | State Committee of Statistics, 2025-01-01 — 12 districts match the published figure | Official |
| Population cross-check | WorldPop 2020 — corroborates the city total to 1.03× | Validated |
| Building heights | GHS-BUILT-H (JRC/Copernicus) — 100 m cell average, not a per-building measurement | Estimated |
| Demand per building | SNiP norms × residents × floor area × hour | Derived |
| Pipe & cable geometry | Laid on the real street network — the utility's own as-builts are not public | Synthetic |
Build the engine once. Every subsystem reads it.
A new utility is a profile and a module — not a new product. Water, electricity and gas were all added without touching the core.
Signals that learn the queue.
Agent-based traffic on the real network, with reinforcement-learned signal control benchmarked against fixed-time and actuated baselines.
The compounding bit
Emergency routing, fire coverage and dispatch all read this same graph at the same congestion factors. The road model is paid for once and used four times.
One demand model. Three utilities.
Water, electricity and gas are all f(residents, building type, hour), derived per building and routed through a network laid on real streets.
Why one model matters
When the district population data was corrected, every utility moved with it the same day — because there is one occupancy estimator, not three.
Coverage is a road problem.
Drive time from the nearest of 19 fire stations to every node in the network — one multi-source shortest path at peak congestion.
One input. Seven systems.
A heatwave is a temperature change. Everything after it is the dependency solver propagating that change until it converges.
- 01 Air-conditioning load rises +35%
- 02 — substations overload and trip
- 03 — water pumps lose power
- 04 — hydrant zones lose pressure
- 05 — hospitals fall back to generators
The cheapest thing that stops it.
— candidate interventions, each applied to the model, re-solved against the same heatwave and ranked by residents protected per million spent.
—
—
What we would need from you.
Real today
Buildings, roads, official population, height priors, derived demand, the dependency graph, and every engine that runs on them. It runs offline, on one machine, from a shipped bundle.
What the city adds
Utility as-builts replace the synthetic pipe and cable geometry with the real network. A telemetry feed turns a static model into a live one. Neither changes the engines — only the quality of what they read.
Everything on the synthetic line is labelled as synthetic in the product itself, not just in this deck.
The people who built it.
Shokiriy ZohirjonMoxirjon o'g'li CEO · Frontend developer
Qodirov AfzalAlijon o'g'li Backend developer Amity University in Tashkent
Shoamir ShorustamovShoakbar o'g'li AI / ML engineer
Give us the as-builts.
We will give you the answer.
A defensible model of the city, before the money is spent.