United KingdomPublic sector2025 – ongoingHM Treasury
HERMeS internal document retrieval assistant
HM Treasury reported around 20-30 weekly users of its internal document chatbot
Problem
Treasury staff searched SharePoint with keywords and read multiple documents to locate relevant information. This made finding internal guidance time-consuming.
Approach
The Treasury's Data Hub built HERMeS using retrieval-augmented generation over internal SharePoint documents. It retrieved relevant passages, generated answers and linked to source documents. Staff had to validate answers against those documents.
Results
Takeaway
HERMeS narrowed document searches and supplied references for checking answers. Reported weekly use did not quantify time saved.
Categories
- Industry
- Public sector
- Business function
- Internal work and knowledge management
- Innovation type
- Process innovation
- AI technology
- Generative AI
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
HM Treasury described an internally built chatbot grounded in SharePoint documents. Around 20-30 staff used it at least weekly, with human validation required for generated answers.
Full details are in the original sources.