FinlandPublic sector2025 – ongoingFinnish Tax Administration (Vero)
AI for document digitisation, tip screening, phone bots and staff assistants
Finland's Tax Administration lists AI tools in use, from reading 2.8 million paper pages a year to phone bots
Problem
The agency receives about 2.8 million pages of paper tax documents a year, and thousands of fraud tips and many phone calls that staff must sort.
Approach
A service provider runs a rule-based AI that digitises the paper documents, with machine learning checking the scanned data and uncertain items going to a person. A model identifies low-risk fraud tips that can be closed, and staff check its decisions regularly. Phone bots route calls, ease congestion on the personal-taxation line and collect data for tax cards. An in-house subtitling tool costs far less per year than outsourcing, the agency says. Staff also use Microsoft Copilot Chat and Copilot Studio assistants that use only guidance sources and exclude customer data.
Results
Takeaway
AI supports document processing and enquiries, while people remain responsible for its output and all tax decisions. The agency has not published measured results.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Innovation type
- Process innovation
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
The agency uses AI to digitise about 2.8 million paper pages a year, screen low-risk fraud tips and assess shadow-economy risks. It also uses phone bots, a chatbot, a subtitling tool and Copilot-based staff assistants. AI makes no tax decisions. The agency has not published results data.
Full details are in the original sources.
