evidoria

← Back to browse

Good practice Imported

MIA — Serbia's Municipal AI Assistant for Citizen Services in Sombor and Šabac

Serbia · Sombor · See the Serbia profile

Evidence: Observational / pre–post Top 66% 53/100 · Ask Evidence Copilot about this practice

Germany's GIZ, Serbian alliance NALED and tech firm SAGA rolled out MIA, an AI chatbot for civil-registry and tax questions in Sombor and Šabac (plus two Bosnian towns) — registering close to 80,000 users and 140,000+ conversations across the four pilots in year one.

~80,000
Registered users (four pilot municipalities combined) (first ~14 months)
>140,000
Conversations logged (four pilot municipalities combined) (first ~14 months)
up to 2,400 hours/year
Estimated staff time on routine queries pre-launch
MIA — Serbia's Municipal AI Assistant for Citizen Services in Sombor and Šabac

Details

Maturity
Pilot
Promoter
City Administrations of Sombor and Šabac, with NALED, GIZ (BMZ develoPPP.de programme) and SAGA
Period
October 2021 – present
Keywords
AI chatbot, municipal government, civil registry, public-private partnership, citizen engagement

Context

In October 2021, the German development agency GIZ — through the Federal Ministry for Economic Cooperation and Development's develoPPP programme — partnered with the Serbian business alliance NALED and technology company SAGA to launch MIA (Municipal Intelligent Assistant), an AI chatbot for citizens' questions about civil-registry certificates, tax certificates, and parental/child benefit procedures.

Objectives

Automate answers to routine, repetitive citizen queries on civil-registry and tax procedures, freeing municipal staff time and improving access to information through channels citizens already use.

Activities

The chatbot went live simultaneously in two Serbian cities, Sombor and Šabac, and two Bosnian municipalities, Bijeljina and Laktaši, reachable through Viber, Facebook Messenger and each city's web portal, under one shared technical platform and governance structure involving GIZ, NALED, SAGA and the Chamber of Commerce of Republika Srpska.

Results

NALED, which coordinated the roll-out, reported that within a little over a year of operation MIA had registered close to 80,000 users across the four pilot municipalities and logged more than 140,000 conversations. Ahead of the launch, NALED had estimated that municipal staff in the pilot cities were together spending up to 2,400 hours a year answering the same routine queries the chatbot was built to handle.

Conclusions

The programme was explicitly designed for replication, running concurrently in two countries and four municipalities from the outset, and organisers stated that positive results were meant to inform wider regional rollout. More than four years after launch, however, available sources show the deployment still limited to the original four pilot municipalities, with no confirmed further expansion, and no published data on response accuracy, escalation rate or a citizen data-protection policy.

Implementation
Indicative cost
Low (< €50k) — Funded through Germany's BMZ/GIZ develoPPP.de public-private co-funding programme alongside NALED and technology partner SAGA; specific budget figures not published.
Time to results
Medium (1–3 years) — Launched October 2021 across four municipalities simultaneously; still operating in the same four municipalities as of the most recent reporting, with no confirmed expansion despite regional-rollout ambitions.
Staffing & skills
GIZ (BMZ develoPPP programme), NALED (coordinating business alliance), SAGA (technology vendor), City administrations of Sombor and Šabac, Chamber of Commerce of Republika Srpska

Conditions for success

  • multi-partner public-private governance structure (GIZ, NALED, SAGA, city administrations)
  • multi-channel access (Viber, Facebook Messenger, web portal)
  • shared technical platform enabling simultaneous multi-municipality rollout

Common failure modes

  • more than four years after launch, deployment remains limited to the original four pilot municipalities despite stated ambitions for regional rollout
  • no published data on response accuracy, escalation rate, or data-protection policy

Where it fits

Governance type
municipal government, donor/public-private partnership
Scale
pilot across four municipalities in two countries
Income level
upper-middle-income

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

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

Attachments

Similar practices you may find useful