AI matching engine for a fragmented connector marketplace
An AI-powered B2B platform that identifies the right connector from millions of spec-driven SKUs – extracting specs from 2D sketches and images with OCR, then ranking matches against a global supplier catalogue.
Challenge
The connector market is highly fragmented, with millions of spec-driven SKUs across global suppliers. Selection depends on precise technical specifications and engineering expertise, and our client – a distributor sitting between buyers and manufacturers – was drowning in manual RFQ cycles.
Buyers routinely submitted 2D sketches instead of part numbers, engineers spent excessive time identifying compatible connectors, and no centralised marketplace existed to match global buyers to verified suppliers.
Approach
We built an AI recognition engine for 2D sketches and images, extracting key specifications using OCR and geometric parsing. A matching algorithm scores candidate connectors and prioritises results against the buyer’s constraints.
The platform is wrapped in a B2B marketplace that connects buyers and verified suppliers, with automated inquiry processing that removes the manual workload from distributors and manufacturers.
Tech stack: PyTorch, TensorFlow, MLflow, LangChain, Snowflake, Apache Airflow, AWS. Delivered by a team of 8 developers, 3 QAs, 2 AI engineers, and 1 PM.
Outcome
Phase 1 shipped in six months with a 60% reduction in manual engineering work across the distributor’s RFQ pipeline.
Connector identification and RFQ turnaround accelerated markedly, with improved accuracy in component selection and shorter supplier engagement cycles – unlocking a new revenue channel on top of the existing distribution business.
Let's build what's next
Share your challenge – AI, data, or infrastructure. We'll scope your project and put the right team on it.