Choosing a running shoe online is hard. The specs on paper don't tell you how a shoe fits or feels, and the real answers are spread across thousands of reviews nobody has time to read.

Fittir is our own running shoe recommendation platform, serving thousands of recommendations a month. Shoppers kept needing answers to specific questions before they'd decide.

What we built

An AI product specialist that reads every review for a shoe, from Fittir, YouTube and Google, and learns what runners consistently say about it. It then answers questions the way a knowledgeable shop assistant would.

Questions answered up front: Each shoe page shows answers to common questions about length, width, feel and fit.

Pros and cons from every review: New reviews are summarised into pros and cons as they arrive, and each shoe's summary reflects the trends across all of them.

Ask anything: Shoppers can chat with the specialist about a shoe, and it answers from the reviews.

Size and width advice: Shoppers say how their current shoe fits and get a size and width recommendation for the new one.

Always current: It checks new reviews every day and updates a shoe's answers when something new comes up.

What changed

Shoppers get fit and feel answers from thousands of real reviews, right on the page where they're deciding. The answers update themselves every day, so nobody has to read and summarise reviews by hand.

Built with

Node.js, JavaScript, OpenAI, Puppeteer and MongoDB. Reviews are collected from the web and processed in a daily job, using a mix of AI models chosen to balance cost against accuracy.

Fittir Running Shoe Specialist AI Agent

What we built

What changed

Built with

Related projects

Tender Writing AI: Lower Costs, Better Answers

Onsen Hot Pools Rostering Automation + AI Agent

Coachi - AI Running Coach

We turn your goals into AI and software that actually works