Item identification, sold comp data, active market velocity, and condition-adjusted pricing — from a single API call. Built for reseller-tools operators and high-volume eBay sellers.
A single photo or keyword query returns a complete intelligence package: item identification via AI vision, median sold prices by condition tier (new, pre-owned, new other), active market velocity as a sell-through rate, price trend direction, and a data confidence score.
Reseller-tools companies need pricing intelligence but lack the infrastructure to build it. Aggregating sold comp data, normalizing it by condition, calculating velocity, and maintaining a cache layer across thousands of items is a significant engineering problem. This API solves it at the infrastructure level so operators don't have to.
Python + FastAPI deployed on Railway. Item identification via Anthropic Claude vision. Image cache built on Supabase with pgvector for perceptual hash similarity matching. Redis for hot-layer caching. Sold comp and active market data currently sourced through authorized third-party data aggregators while direct platform API access is pursued.
A perceptual hash cache middleware layer sits between the vision pipeline and the data layer. Items are fingerprinted on first identification using pHash — a 64-dimension vector stored in Supabase with pgvector for similarity matching at a 0.85 cosine threshold. Repeat lookups on cached items resolve in under 500ms without touching the Vision API. Cache density compounds across all deployments — every new item identified by any operator enriches the shared cache, reducing costs and latency for every subsequent lookup on that item across the network.
The data layer is built around a source-agnostic abstraction interface. Data sources are swappable without modifying the pipeline above them. The normalized output schema — median price, velocity badge, price trend, condition distribution, sample size, and confidence score — stays constant regardless of source. This architecture was designed from the outset with direct eBay API integration as the target end state. Acine is actively pursuing access to eBay's Developer Program to replace third-party aggregation with direct platform data, which will improve data freshness, accuracy, and reliability for all downstream operators and users.