Mohit Kumar

FindBack

Flip the Switch Hackathon/Nov 2026

Getting a lost item back should take more than being first to say, “That looks like mine.” FindBack helps students and campus communities recover lost belongings through the chat they already use. A common black headphone case can match several people’s descriptions, while sharing its identifying details publicly gives everyone the answers. Finders, claimants and library staff are left carrying context between conversations. FindBack coordinates that handoff inside Mattermost. It responds to found-item conversations and guides an enrolled claimant into a private recovery chat. An Intake agent helps the claimant confirm their report and describe identifying details in their own words. A Matcher agent prepares a staff-only comparison against the library’s observations, separating agreements, contradictions and missing information. The librarian decides whether to approve the exact claim and item. FindBack then reserves it and delivers collection instructions privately. The video ends with approval and reservation; a completed return additionally requires staff handover and claimant receipt. People contribute knowledge and judgment; agents keep the process moving. The demonstration uses fictional participants and objects, with real Switch transport, Mattermost messages and Codex inference.