Bartcash

Scout: an AI barter agent built on Qloo’s taste graph. It finds who has what you want, who’d take what you have, and drafts a fair offer. Swap what you have for what you want.

Comment

Inspiration

A lot of value sits in things people own but no longer use, and in places like Nigeria cash is often the scarce part of a trade. Bartcash is for anyone who wants to buy and sell but doesn't have the cash to: they can trade what they own for what they need. Barter fixes the cash problem but has a discovery problem: you need someone who has what you want and wants what you have. I started building Bartcash as a startup in September 2026. When I heard about the Qloo hackathon, I saw that taste is exactly the missing signal: people who like the same things tend to want each other's things.

What it does

Bartcash is a mobile marketplace for item-for-item swaps, with an optional cash top-up when values don't line up. Scout is the trading agent inside it:

  1. Reads your listing. It recognises the brand in the title and description and looks it up in Qloo.
  2. Suggests what to ask for. Qloo's related brands are mapped onto Bartcash's categories, so Scout suggests five categories to ask for in exchange, with a reason ("Fans of Canon also like Sony").
  3. Finds and ranks matches. For every listing, Scout scores other people's listings on four signals: does it fit what you asked for; does that person want what you have; taste affinity (one Qloo insights call scores the candidates' brands against your listing's brand and your interests); and how much your tastes overlap. Matches are labelled Strong / Good / Possible, with "Why this matches".
  4. Finds people like you. You add brands, artists, films, games and more to a taste profile (Qloo search). Scout compares profiles and shows people with similar taste, what you have in common, and their listings that fit you.
  5. Drafts the offer. One tap: Scout picks your listing to offer, suggests a fair cash top-up from the two automated valuations, and writes an opening message that mentions an interest you both share. You review and send every offer yourself; Scout never sends anything without you.

Turn Qloo off and Bartcash has no matching at all: the agent only exists because of Qloo's taste graph.

How I built it

  • App: Expo / React Native (Android), Expo Router.
  • API and worker: Node.js, Express 5, TypeScript, Prisma, PostgreSQL (Supabase), pg-boss job queue. Images on Cloudflare R2. Hosted on Render.
  • Qloo: search (all entity types), insights (related brands, taste affinity), tags, popular entities. A client with caching, request pacing, retries, a circuit breaker and a monthly budget guard; every Qloo feature sits behind one flag.
  • Background jobs: brand lookup and want suggestions per listing, match refresh per listing (one Qloo insights call), person-to-person taste compatibility, automated valuation from the cover photo.

Challenges I ran into

  • Turning Qloo's brand-to-brand affinity into "what would this person swap for": mapping related brands onto categories with position decay so that, for example, car makers that appear deep in every popular brand's list don't swamp the suggestions.
  • Making matches explainable: every score is stored with its reasons, so the app can show "Why this matches" without another call.
  • Running on free tiers (Render, Supabase, Groq) without breaking under load: connection-pool limits, rate-limit-aware job retries, and a server wake-up flow in the app.

Accomplishments that I'm proud of

  • An end-to-end agentic loop (listing → suggestions → matches → offer) where every step is grounded in Qloo data and explained to the user.
  • Taste-based matching that still respects what people actually want to trade.

What I learned

Taste is a strong predictor of trades, but only alongside explicit intent; the best matches combine "you asked for this", "they want yours" and "you like the same things".

What's next for Bartcash

Launch with real users, a web version, push notifications for new strong matches, and Scout suggesting the best swap across all your listings at once. I'm looking for funding to take Bartcash from hackathon build to launch.

Built since Sep 30, 2026

The marketplace (accounts, listings, valuation, trades, chat, ratings) existed before. Added for this hackathon: the whole Qloo integration and Scout (brand lookup, want suggestions, matching with taste affinity, taste profiles, people like you, Draft an offer), Home / Discover / taste screens, in-app tours and guides, listing photo editing, Inbox improvements, deployment hardening.

Share this project:

Updates

Submission history