Skip to content
← All products
AI Commerce Beta Cartway

Merchandising and conversational commerce that reads intent

For mid-market merchants whose search fails on the descriptive and comparative queries that convert best. Ranks by intent against structured catalogue attributes, and answers product questions only from merchant data, with a citation to the source field.

Built with a merchant during a forward-deployed engagement. In beta with a small number of mid-market merchants.

+31% Conversion lift, median deployment
2.1M Shopper queries resolved monthly
94% Answers grounded in merchant data

The problem

Storefront search still fails on the queries that convert best, namely the descriptive, comparative, and constraint-shaped ones. Shoppers who cannot find the product leave, and the analytics record it as a bounce rather than a miss.

Our approach

We build a structured representation of the catalogue, then resolve queries against attributes and constraints instead of embeddings alone. Product answers are grounded strictly in merchant data with citations back to the source field, so a wrong spec is traceable rather than mysterious.

It is in beta with a small number of mid-market merchants.

Why grounding matters commercially

A hallucinated product specification is not a model quality problem, it is a returns problem and in some categories a compliance one. It refuses to answer rather than guess, and the refusal rate is a tracked metric rather than a hidden failure.

Working on something like this?

We take on a small number of engagements at a time. Bring the problem.

Free consultation