evidoria

← Back to browse

Good practice Imported

BolognAI — Bologna's Human-in-the-Loop Framework for Generative AI in Municipal Administration

Italy · Bologna · See the Italy profile · See the Bologna profile

Evidence: Descriptive / self-reported Top 35% 67/100 · Ask Evidence Copilot about this practice

After a 10-month, 115-staff needs assessment identifying 279 AI use cases, Bologna's council approved an AI governance model in Dec 2025 and launched #BolognAI in Feb 2026, using Gemini and NotebookLM so staff draft and review administrative acts under mandatory human sign-off.

115
Staff interviewed in needs assessment
279
AI use cases identified
10 months
Programme preparation duration
more than 50
Interview hours conducted
BolognAI — Bologna's Human-in-the-Loop Framework for Generative AI in Municipal Administration

Details

Maturity
Pilot
Promoter
Comune di Bologna (Municipality of Bologna)
Period
2025-2026
Region (NUTS)
ITH55
Keywords
public administration, generative AI, digital transformation

Context

The Municipality of Bologna ran a ten-month programme, structured as three 'Cantieri dell'Innovazione' (innovation work-streams) covering governance and risk management, needs assessment, and change management, to prepare a formal framework for generative AI in city administration.

Activities

More than 50 hours of interviews with roughly 115 staff across the municipality and the metropolitan authority surfaced 279 potential AI use cases, from which the city is selecting projects with the highest impact and lowest risk. The city council approved the governance model in December 2025, aligning it with the EU AI Act, and the programme -- branded #BolognAI / 'Intelligenza Urbana' -- officially launched on 28 February 2026. Under it, administrative staff use Google Workspace's Gemini and NotebookLM as drafting and review 'agents' to check the completeness of resolutions and determinations, revise text for gender-neutral language, and improve accessibility. Every administrative act produced with AI assistance carries a mandatory notation disclosing that assistance, and legal validation and final responsibility remain explicitly with the human official.

Results

Only process metrics are public so far: 115 staff interviewed and 279 use cases mapped. Because #BolognAI launched only in February 2026, no outcome data -- time saved, error rates, or citizen-facing effects -- has yet been published.

Conclusions

The public evidence available so far documents process design, staff engagement and governance approval, not measured impact; whether the programme delivers efficiency or quality gains once it is operating remains to be demonstrated.

Implementation
Indicative cost
Low (< €50k)
Time to results
Medium (1–3 years) — Ten-month preparation programme through 2025; governance model approved December 2025; #BolognAI launched 28 February 2026.
Staffing & skills
Comune di Bologna (Municipality of Bologna) administrative staff, Bologna metropolitan authority staff, named responsible officers within the three Cantieri dell'Innovazione work-streams

Conditions for success

  • structured, staff-driven needs assessment before technology selection
  • formal governance model aligned with the EU AI Act
  • mandatory human sign-off and disclosure notation on every AI-assisted act

Common failure modes

  • no outcome data (time saved, error rates, citizen effects) published yet as of the February 2026 launch

Commonly funded by

National / regional programmes Recovery and Resilience Facility (national plans)

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

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