Contact us
Challenge

How might we responsibly use Agentic AI to help municipal technicians perform the preliminary analysis of building-procedure applications, making the process faster, more consistent, more transparent and easier to manage, while preserving human oversight and administrative accountability?

Comune di Padova wants to improve the preliminary review of building-procedure applications, such as SCIA, CILA and building-permit requests, using responsible Agentic AI. The challenge aims to test an AI-assisted, human-in-the-loop workflow that helps municipal technicians extract information, check completeness, connect evidence to applicable rules and produce traceable preliminary reports without replacing human responsibility.

Comune di Padova

Comune di Padova is the municipal authority governing Padua, a major city in Italy’s Veneto region. It manages essential public services and supports the city’s social, cultural, environmental and urban development, working to improve quality of life and deliver more accessible, efficient services for residents and visitors.

Read more

Challenge framing

Current Situation

Building practices include heterogeneous administrative forms, technical reports, professional declarations, plans, drawings and other graphical materials. These files must be reviewed against national, regional and municipal rules, as well as planning instruments.

Today, submitted files are manually reviewed by municipal staff, who read documents, check plans and maps, verify consistency and request missing or corrected material when needed. The current workflow relies on “Impresa in un giorno”, the municipal Protocol system and Lizard GPE.

Current pain points include time-consuming manual reading, uneven review times, heterogeneous documentation, complex regulatory interpretation, dependency on individual expertise, risk of errors or inefficiencies, integration requests and difficulty tracing the full process.

Desired Situation

Comune di Padova wants a human-in-the-loop building-compliance assistant that supports technicians in analyzing practices in a more structured, consistent and traceable way.

The solution should ingest selected documents and drawings, extract relevant information, compare extracted data with applicable regulations or checklists, identify missing or inconsistent information and generate preliminary reports for human review.

The assistant should act as a support layer, not as the decision-maker. Formal administrative responsibility remains with the municipality and competent technicians, with AI outputs limited to recommendations, checks, evidence summaries and decision-support signals.

Stakeholders

Stakeholders

Internal stakeholders
Municipal technicians, building-procedure officers, Edilizia Privata Department, SUE Office, Settore Innovazione e Transizione Digitale, IT/security teams, data owners, DPO/privacy, legal/compliance, internal control and municipal managers.

External stakeholders
External professionals such as architects, engineers, surveyors and technical consultants, citizens, businesses, startups, solution providers, NTT DATA, universities or research partners where relevant for validation.

Primary users
Municipal technicians, building-practice officers and SUE-related users involved in preliminary review and formal approval of building-procedure requests.

Possible solutions

Possible Solutions

  • Agentic AI assistants for regulated workflows.
  • Document intelligence platforms for technical and administrative files.
  • Multimodal AI for PDF, CAD, plans, scans and mixed document packages.
  • Compliance-checking and regulatory intelligence tools.
  • RAG and knowledge-base solutions for legal or technical corpora.
  • Workflow automation tools for case review and report generation.
  • Explainability, auditability and evidence-traceability layers for AI outputs.
  • AI governance, model monitoring and human-review workflows.
  • Geospatial intelligence and planning-data integration capabilities.

Technical Requirements

  • Ingestion and organization of heterogeneous building-procedure documentation.
  • Analysis of Italian technical reports, declarations, forms and textual documents.
  • Interpretation support for plans, sections, elevations, CAD/PDF files and scanned drawings.
  • Extraction of technical data such as surfaces, volumes, intended use, distances, heights and accessibility clearances.
  • Detection of missing documents, missing fields, inconsistencies or contradictions.
  • Retrieval and reference of relevant national, regional and municipal rules.
  • Structured preliminary reports with evidence, source references, confidence levels and uncertainty flags.
  • Human review, correction, approval, rejection and feedback loops.
  • Audit trail for inputs, outputs, model actions, user actions and report versions.
  • GDPR, privacy-by-design, role-based access, secure authentication and alignment with public-administration AI guidance.
Apply now
Apply now

Copyright © 2024 NTT DATA, Inc.