# Lisovate

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Pawel Lisowski](https://cerebralvalley.ai/u/Paweljlisowski)
- **GitHub:** https://github.com/Lisovate/plomba
- **Demo video:** https://www.youtube.com/watch?v=DokZylSNUKc
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3772

I live in Gdynia, a port city in Poland. A container that leaves the port here changes hands five or six times before it gets to the customer: terminal, train, inland terminal, truck, depot. Every time it changes hands, the liability for damage moves too. But the condition is still written on paper or shot as a few random phone photos, so when a dent shows up at the end, nobody can prove where it happened and usually the last company in the chain pays.

Plomba is my attempt to fix that. It's a web app the checker opens on their phone. They point it at the doors and it reads the container number (and checks it against the ISO 6346 check digit, so a misread can't slip through). Then they film one walk around the container. Plomba tracks the box through the video, works out which side is which, and maps every photo onto the container itself in metres. That's the important bit: because the damage is stored on the container and not on the photo, the next checker in another city sees the old damage drawn on exactly the same spot, on a new photo or even on the live camera. They just tap "still there" and only record what's new.

New damage can be found by the model or simply said out loud in Polish. Each handover gets sealed into a hash chain, so nobody can quietly edit it later. From that, the container's damage passport shows between which two handovers each dent appeared, and the app can print the receipt, send the standard EDIFACT gate message and draft a claim.

Live demo: https://app.plomba.lisovate.com (access code: plomba-f12093c2). The quickest way in is the Try page, which plays real container footage with the AR on it, no container needed. Then open container PLBU 400004 3.
Website: https://plomba.lisovate.com

It's open source (Apache-2.0) and runs in one Docker container, and there's a waitlist for a hosted version. I built it during the hackathon; the build log is in the repo.

## More from Meta Global AI Developer Hackathon

- [FollowAlong](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3769)
- [Ripple handoff](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3770)
- [BOLUS](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3771)
- [Abra](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3773)
- [Show Me](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3774)
- [Prysm](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3775)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/3772. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt