Investment fund · private equity, NDA
AI pilot: analytics in hours instead of weeks
- Context
- An investment fund. Preparing analytics on portfolio companies and markets took the team weeks: data scattered, collection manual, every analyst request worth days of work.
- Task
- Integrate Gen AI into core business processes so that data stays under the fund’s control and the effect is measurable — not "AI for AI’s sake".
- What was done
- Designed and launched a Gen AI / LLM / MCP integration pilot across business processes. Built hybrid infrastructure (on-premise + Yandex Cloud + AWS) respecting data sovereignty requirements. Assembled a cross-functional team and introduced Agile/Scrum with DevOps elements.
- Result
- The analytics prep cycle shrank from weeks to hours. The "LLM-first prototyping" practice — a fast PoC on real data before scaling — became the fund’s standard.
- Stack
- LLM, RAG, MCP, on-premise + Yandex Cloud + AWS, Agile/DevOps