AI matching engine for a two-sided marketplace

Challenge Our client is building a marketplace that connects climate technology companies with corporations and governments working toward sustainability and net-zero targets, across renewable energy, biodiversity, circular economy, agriculture, and food systems. Buyers need to find solution providers aligned to specific initiatives and targets; providers need visibility with the right buyers. When they came to us they had a clear idea and no product. They needed to go from zero to a launched MVP: a web platform and a matching engine sitting at the core of it. The central technical challenge was that the product's value is not directory search. It is intelligent matching - reasoning over deep vertical data to surface the most relevant providers for each buyer's specific goals. That requires an AI layer on top of the platform, not keyword filters or faceted search. The product had to serve both sides: providers uploading and maintaining company profiles, and corporate and government buyers being matched against their sustainability initiatives and targets. Delivering that in a single MVP, with the matching engine as the differentiator, was the brief.

Approach We took the product from zero to launch - no existing codebase, no prior product. The engagement was scoped as a full MVP build: define the platform architecture, build the web application, and implement the AI matching engine that would separate the product from a static directory. The decision to add an AI layer was driven by the proposition, not by fashion. Buyers have specific initiatives, timelines, and vertical focus areas. Providers have detailed capabilities across sectors and solution types. Matching the two required reasoning over structured and unstructured data - identifying fit by initiative, sector, and capability rather than by keyword overlap. The layer was designed to consume profile and initiative data, analyse it, and return ranked, relevant matches. Technical choices favoured speed to launch and clarity of ownership: a React single-page application for the product surface, and the matching engine built as a dedicated backend component fed by the same data that powers the platform, so profile updates flow through to match quality without a manual sync step.

Solution We delivered the full MVP: the web platform and the AI matching engine. React web application The product surface is a React web application providing the primary experience for both solution providers and corporate or government buyers. Company profile management Solution providers create and maintain company profiles covering solution areas, sectors, capabilities, and the other attributes the matching engine draws on to find relevant buyers. AI matching engine The matching engine is the core technical component. It sits on top of the platform data and uses an AI layer to analyse buyer initiatives and targets against provider profiles, returning ranked matches so buyers see the most relevant providers for their specific goals rather than a flat list or keyword-only results. Two-sided matching Buyers specify their sustainability initiatives and targets, and the engine surfaces the most relevant providers. The result is a genuine two-sided marketplace: providers get discovered by the right buyers, buyers get a shortlist that fits their initiatives.

Results The MVP launched. The web application is live, providers onboard and maintain profiles, and buyers are matched against their initiatives and targets. The AI matching engine is in production and drives those results. The client has a shipped product delivering intelligent matching rather than directory-style search, on a single codebase with a clear path to iterate on both the platform and the matching logic. What this engagement demonstrates: we can take a founder from a blank page to a launched product with the hard part - the reasoning layer that makes the product worth using - built in from the start rather than deferred to a later phase.