Case Study - Understand Gamer Conversations at Scale
ThoughtRay is a highly ambitious initiative by industry leaders in AAA gaming that seeks to revolutionize how studios and publishers interact with gamers.
- Client
- ThoughtRay
- Year
- Service
- Product Design, Web Development, Data Engineering, Artificial Intelligence, R&D
Overview
TRAILR is an AI-powered deep listening platform that helps the gaming industry deeply understand gamer responses to their digital assets.
We led the development process, starting from zero, providing product guidance and technical execution on a complex product where AI permeates the user experience in almost every element.
Customer feedback data is notoriously hard to use, but unlocking it has significant benefits - it lets marketers measure real customer response in minute detail, in real-time, in-market. To make this feasible, we developed a new kind of large language model called context injection, fine tuned for the gaming industry. We worked on language identification, sentiment and emotion, topic extraction, statement splitting, clustering, semantic search, and both extractive and abstractive summarization, blended in a beautiful, fast and functional UI.
On the data side, we built efficient pipelines that collect and process millions of pieces of feedback, index new videos, resolve game references, and align databases to allow for a single market-wide view of a game's potential, served via a REST API.
We believed enough in the promise that we took an equity position in the business, which now serves some of the biggest names in gaming.
What we did
- Frontend (Next.js)
- Infrastructure
- Large Language Models
- Semantic Search
- Retrieval-Augmented Generation
- Machine Learning / AI
- Data Engineering
- Development time
- 9 months
- Videos tracked
- 45,000
- Comments processed
- 15M