Conversational product search for a global sportswear e-commerce store
A global sportswear brand wanted to know whether shoppers would use natural language to find products, without rebuilding its highest-converting pages. In under six weeks, three engineers and a designer built a conversational layer on top of the existing product page, trained on 500-plus products, and tested it with real shoppers. Shoppers explored more products and reported more confidence, and the brand added the concept to its roadmap.
- 6 weeksto a deployed, user-tested prototype
- 500+products in the trained catalogue
- 4people on the build — three engineers, one designer
- Client
- Global sportswear brand
- Industry
- Retail & consumer
- Practices
- AI & Agents
- Timeline
- Under 6 weeks
- Stack
- Pinecone, LangGraph, Next.js, Node.js
Shoppers describe. Dropdowns cannot listen.
Twistag built a conversational product discovery layer for a global sportswear brand — natural-language search embedded directly in the existing product detail page, trained on a catalogue of 500-plus products, deployed and user-tested with real shoppers in under six weeks by three engineers and a designer. The brand's shoppers could describe what they wanted ("something bold and minimal for a city weekend") but could only express it through a colour dropdown. Twistag built a headless component that layers over the existing PDP without touching the e-commerce platform, using Pinecone for semantic product matching at sub-500ms latency and LangGraph for conversation state. The brand is not named publicly.
Product detail pages have looked the same since the early 2010s — hero image, tabs, size selector, filters in a sidebar — while the way people interact with software has changed completely. Shoppers ask questions. They describe what they want in natural language and expect the system to understand intent. A shopper looking for "something bold and minimal for a city weekend" had no way to express that through a colour dropdown. The commercial constraint made experimenting hard: product pages are the highest-converting pages on any e-commerce site, which makes them the riskiest place to change anything. The brand's digital team did not want a PDP rebuild — they wanted to know whether conversational commerce could be validated quickly and cheaply, with enough realism to produce behavioural signal from real shoppers.
A layer on the page, nothing rebuilt.
On top of the page.
The conversation is an optional layer over the existing product page, so shoppers who prefer browsing are unaffected and the e-commerce platform is untouched.
Search by meaning.
Pinecone stores every product as a vector and returns the closest matches to a description in under 500 ms, while LangGraph keeps track of the conversation for follow-ups and comparisons.
A knowledge layer for fashion.
Shopper language is translated into product attributes: “casual but not boring” maps to style characteristics, “gym to dinner” narrows differently from “weekend travel”, and the tone matches the brand.
Six weeks, weekly demos.
Discovery in week one, build with weekly demos in weeks two to five, and testing with real shoppers in week six.
Validated with real shoppers in six weeks.
Twistag delivered a functional prototype in under six weeks with three engineers and one designer. User testing with real shoppers confirmed the hypothesis: shoppers engaged more deeply, explored more products, and reported higher confidence in their selections. The brand advanced the concept to its production roadmap. Because the component is headless, a production rollout can proceed page by page or per A/B segment without a PDP rebuild. These are process and qualitative outcomes, stated as such — this case study reports no conversion, revenue or retention figure, because none has been published by the brand.
The engineering value lies in the knowledge layer that translates between user intent and the brand's product reality.
- Pinecone
- LangGraph
- Next.js
- Node.js
Twistag built a conversational product discovery layer over a global sportswear brand's existing product detail page, trained on 500-plus products.
