The service helps people choose a product and saves time on searching for offers. The shopper writes or says in their own words what they are looking for ("I want a laptop"), and the AI advisor replies with a selection of suitable offers with links to stores. The service collects products from online stores. We took part in developing part of the project.

Challenge

Replace filter-based search with conversation: the shopper describes the need as they would to a store clerk, and the service picks store offers by itself.

What we did

  1. AI advisor

    A chat that understands a request in plain language and replies with a selection of store offers. It runs on an LLM.

  2. Voice input

    A request can be dictated.

  3. Store aggregation

    Offers from online stores are collected into one catalog: more than ten sections, from home appliances and electronics to kids' goods and sports.

  4. City selection

    Offers are shown for the selected city.

  5. Favorites, comparison, and a personal account

    Shoppers can save products and compare them.

  6. Analytics on ClickHouse.

Screenshot of AI shopping advisor (#2)
AI shopping advisor: screen 2
Screenshot of AI shopping advisor (#3)
AI shopping advisor: screen 3

Stack

Node.js, PHP, PostgreSQL, Supabase, MySQL, ClickHouse, Redis, LLM, ML.

Result

The service is currently offline.