All work02
Teams-BOT
A multi-agent AI bot living inside Microsoft Teams, answering equipment questions from a database of 2,395+ construction machines.
- Year
- 2025
- Role
- AI Developer
- Client
- RÜKO GmbH Baumaschinen
- Stack
- Python · Microsoft Teams API · Multi-agent AI · Vector DB · Cloud-native
01 — Problem
Equipment data for thousands of construction machines sat in a database that most employees never touched directly. Looking up a machine's details meant switching tools, knowing the schema, or asking someone who did.
The audience already lived in Microsoft Teams all day — the answer had to come to them, not the other way around.
02 — Approach
I built a Python bot that plugs directly into the Microsoft Teams API, so asking about a machine is as easy as messaging a colleague.
Under the hood, multiple specialized agents run in parallel — one resolves what machine is meant, others fetch and compose the answer — which keeps responses fast even when a question touches several data sources.
The system is deployed cloud-native and indexes 2,395+ equipment records.
03 — Result
Equipment knowledge that used to require database access is now one Teams message away. The bot answers from 2,395+ indexed records, with parallel agent execution keeping latency low.
