Germany's research ministry deploys a chatbot that answers project-funding rule questions
Staff at the German Federal Ministry of Research, Technology and Space who handle project funding must follow a 200-page Project Funding Manual that was hard to search on the intranet. In six months a team of the ministry, deepset, HalloWelt, GovTech Campus and the Fraunhofer Institute built a chatbot that answers questions about project funding rules in natural language.
- Agent or tool
- Haystack (deepset)
- Topic
- Knowledge & Notes
- Reported on
- Case study (deepset)
- Added
- 2026-11-13
Why it matters
Notes are only useful if you can find and connect them later. Agents that organise, summarise and link information make a personal knowledge base far easier to keep up. It is one of 62 sightings filed under Knowledge & Notes on Agent Sightings (number 9 by interest), and one of 3 involving Haystack (deepset).
How you could try it
Here is a reusable prompt, written by us in our own words, that captures the task. Paste it into Haystack (deepset) or any similar agent and adjust the details to your situation.
I want an assistant that answers staff questions from our own documents and cites where each answer came from. The tool in the source case was Haystack (deepset).
Tip: Start small: run it once on a low-stakes example, read what the agent did, and only then let it run on real work.
Source
Read the original first-hand account on Case study (deepset): https://deepset.ai/case-studies/german-federal-ministry-research-technology-spa…. Our summary is a paraphrase; details are in the source.
Featured in these guides
Built an AI agent? Feature your agent on Agent Sightings, free submission or a featured spot.
Browse more like this
Related sightings
- Bosch built DRIVE Companion, an AI assistant used by 600+ people on SAP migration teams
- Paramount Residential Mortgage Group answers employee questions with Capacity team AI support
- NIH runs an air-gapped h2oGPTe assistant for 8,000 employees and deflects about 10,000 IT tickets a year
- Run a nutrition bot for meal decisions and calorie logging
- YPulse moves from fixed RAG pipelines to AI agents for its enterprise research customers