Raft's AI4BI: An AI Analyst for Coaches and Scouts in Soccer, Hockey and Basketball

In 2026, Raft described the development path of AI4BI, an artificial intelligence platform for sports analytics. Built in 2017 for cricket, it has grown over nine years into a universal tool for clubs in five sports and for the mass-market fitness segment.

How AI4BI reached its current scale

Raft built the first AI4BI prototype in 2017 specifically for cricket, a sport that is practically exotic on the Russian market. In the team's own words, the solution looked like "some kind of spaceship" at the time: the market was not yet used to AI analytics in sports.

Over nine years, the platform has grown from a single discipline into a comprehensive product:

• 2017 — launch of the first prototype, sport: cricket
• 2026 — support for soccer, hockey, volleyball, basketball and bandy
• RAG knowledge bases for plugging in domain expertise without retraining the model
• An AI/ML module for forecasting, pattern analysis and recommendations
• A new branch: "civilian" fitness analytics for the non-professional market

Which roles AI4BI is built for

AI4BI is built not as a single universal dashboard but as a role-based system: each user gets their own data layer and workflow logic. Coaches see training loads and tactical patterns, scouts get tools for comparing candidates, and sports analysts get deep breakdowns of matches. Sporting directors and club owners work with a financial and strategic picture tied to results.

The key technical feature is the RAG mechanism. It lets users upload information specific to a club or league — tactical regulations, historical matches, medical protocols — and the model starts answering questions with that context in mind, without retraining. According to Raft, this lowers the barrier to adoption: domain expertise can be plugged in without involving ML engineers.

"Today it is an accessible tool for sports organizations, from professional teams to fitness clubs,"

Raft notes in its corporate blog.

Three cases, three sports, three roles

The central part of the AI4BI presentation is three demonstration use cases, each covering a separate sport and a separate user role. This pairing shows how much the platform's interface and logic change with context: analytics for a hockey scout is built around fundamentally different metrics than the toolkit for a volleyball coach.

The AI4BI fitness branch runs on the same architectural foundation but with different data: not in-game match statistics, but biometric indicators and each client's individual progress over time. A fitness club trainer gets recommendations for adjusting programs, and the client gets a visualization of their own progress.

What this means

AI4BI is a practical example of "domainization" of AI through RAG: instead of a separately trained sports LLM, it uses a universal AI engine enriched with subject-matter knowledge bases. This approach scales horizontally — from cricket in 2017 to any sport and fitness in 2026 — without the cost of retraining for each discipline.