evidoria

← Back to browse

Good practice

FURS AI Chatbot Family — Slovenia's Generative-AI Assistants for Tax, Customs and Duties

Slovenia · Ljubljana · See the Slovenia profile

Slovenia's tax authority (FURS) expanded a generative-AI chatbot — first launched for VAT in May 2025, then income tax, customs, excise and other duties — to about 800 users and 3,000+ questions a day, answering over 800,000 questions in its first six months.

800 users/day
Daily active users (~March 2026 (roughly six months into the newest chatbots' operation))
3,000+ questions/day
Daily questions answered (~March 2026)
800,000+ questions
Total questions answered since rollout (May 2025–~March 2026)
FURS AI Chatbot Family — Slovenia's Generative-AI Assistants for Tax, Customs and Duties

Details

Maturity
Scaling
Promoter
Financial Administration of the Republic of Slovenia (FURS)
Period
2025–
Keywords
taxation, customs, e-government, generative AI, citizen services

Context

Slovenia's tax authority, the Financial Administration of the Republic of Slovenia (FURS), launched a generative-AI chatbot on 8 May 2025 to answer public questions about VAT (DDV) and tax cash registers. Over the following months FURS extended the same approach to further subject-specific chatbots covering personal income tax, business-entity taxes, customs, environmental charges, excise duties and the EU's carbon border adjustment mechanism (CBAM). What began as a single pilot grew into a family of generative-AI assistants covering most of the tax and customs domains FURS administers.

Objectives

The chatbots aim to give the public instant, 24/7 answers to routine tax, customs and duty questions without needing to contact a human agent. FURS also intended the bots to reduce the load on its human call-centre staff so they can focus on complex, case-specific enquiries.

Activities

Content shown by the chatbots is refreshed daily by trained contact-centre staff, and user ratings are fed back into ongoing improvements. By design the bots draw only on FURS's public website and guidance rather than any personal or confidential taxpayer data, and anyone with a case-specific question is directed to the human VAT Call Centre.

Results

By around March 2026, roughly six months into the newest chatbots' operation, Slovenian media reported the chatbot family was fielding about 800 different users and more than 3,000 questions a day, concentrated on weekdays, and had answered over 800,000 questions in total since the rollout began in May 2025. FURS says the bots now field more queries than its human call-centre staff. These usage figures come from Slovenian media reporting rather than an official, continuously published FURS statistics page, so they are credible but not independently audited.

Conclusions

The rapid expansion from a single VAT chatbot to a family spanning income tax, customs, excise duties and CBAM suggests the approach has proven internally replicable across tax domains within FURS. No other country's tax authority has yet been documented adopting the same approach.

Implementation

Indicative cost
Medium (€50k–€500k)
Time to results
Medium (1–3 years)
Staffing & skills
Trained contact-centre staff who refresh chatbot content daily, Human VAT Call Centre staff who handle case-specific or complex queries escalated from the chatbots

Conditions for success

  • Restricting chatbot answers to FURS's public website content and guidance rather than personal or confidential taxpayer data
  • Daily content refresh cycle to keep answers aligned with current guidance
  • Feeding user ratings back into ongoing improvement of chatbot answers
  • A clear escalation path to human call-centre staff for case-specific questions

Where it fits

Governance type
national tax and customs authority
Scale
national
Income level
high-income EU member state

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful