Government as Code

Inspiration

https://www.loom.com/share/db54ce52bbe444caadcdab338bfb0362?sid=401ef836-a6f9-4a2c-ab27-1e33c4ca1cf2 (demo!!) The UK government ploughs money into public infrastructure by repeatedly sending contractors to do the same job (i.e. surveying schools) and ultimately spending multiples more than necessary - whilst still seeing infrastructure break (see the below article) https://www.independent.co.uk/news/uk/home-news/children-injured-ceiling-collapse-thurlow-dulwich-b2401279.html

What it does

We built an end-to-end cockpit to allow government operators to track crowdsourced infrastructure failures (from potholes to critical power grid damage) via social media (Telegram for now), use AI to categorise problems & communicate directly with contractors to gather RFQs/ dispatch labour.

How we built it

Using Telegram Bots, Multi-Modal AI to triage images using government defined rubrics and Palantir foundry to handle data flow, dashboards and orchestration

Challenges we ran into

Learning streaming in Foundry, sourcing high quality data for assets and setting up a flexible ontology.

Accomplishments that we're proud of

Getting a E2E working MVP working in less than 24 hours - collecting real images and locations from near the venue and seeing the data flow through into our platform, and being able to send messages from the platform.