Team 79

Snapback

Snapback is a personal AI agent that handles everything after you buy something: it files returns, warranty claims and price drop refunds, buys working replacements within rules you set, and backs every action with cited store policies and a tamper evident audit trail.

Check your own project

Community votes recognize the build and do not affect official judging.

Review the project

Start with the source code, then open the demo or video if available.

View GitHub repository
Project description
Snap a photo of a broken item, or let it watch your orders in the background. Snapback matches the photo to your order, assesses the damage, and retrieves the store’s real policy with citations. Deterministic rules decide what’s allowed. The agent files the return in a real browser, orders a better rated replacement within your budget, adds a drop off reminder to your calendar, and confirms the refund. You set standing rules once. It acts on its own inside them and asks for a one-tap verified approval only for choices that are yours, like a major price drop. Every step is hash chained, so any edit to the record is detectable. Senso: trusted store policies; every decision cites the exact clause Superlinked SIE: image embeddings for photo matching, order extraction, replacement reranking SingleStore: orders, vector search, tasks, approvals and the hash-chained audit log Persona: verifies the human behind the standing rules and approvals
Tools used