Applied AI · 2026
RAG at Boticário
At Grupo Boticário I design a RAG pipeline over support materials, trainings, and operational guidance, indexed for grounded retrieval. I was also selected to multiply AI across the company: training other teams, coaching implementation, and unblocking adoption.
- Role
- AI / GenAI Engineer
- Context
- Grupo Boticário
- Year
- 2026
- Stack
- RAG · LangChain · Vector store · Agents
The problem
Operational knowledge lived across documents, trainings, and day-to-day guidance. Without grounded retrieval, people either lose time searching or get an answer they cannot check.
What I did
I design the RAG on that material, with a vector index. On the agent side, I implement database lookup, ticket actions, and Slack messaging inside a documented harness so orchestration stays repeatable.
What holds the case up
- 01
Grounded retrieval
The pipeline covers support materials, trainings, and operational guidance. The answer comes from what was indexed.
- 02
Tool-using agents
Database lookup, ticket-board actions, and Slack messages, inside a documented harness.
- 03
AI multiplier
Trainings for other teams, implementation coaching, and support when adoption stalls mid-delivery.