The OECD.AI Policy Navigator

Our policy navigator is a living repository from more than 80 jurisdictions and organisations. Use the filters to browse initiatives and find what you are looking for.

Camilla - Digital Assistant for Public Sector Competitions


Added by:   OECD analyst
Added on:   23 Sep 2026
Updated by:   OECD analyst
Updated on:   23 Sep 2026

undefined
Camilla is a generative AI virtual assistant that helps citizens access clear, up-to-date information on public recruitment competitions through the inPA Portal, Italy’s single gateway for public administration hiring. Developed by Formez PA with CSI Piemonte, it enhances access to notices, simplifies participation, and reduces the workload of staff managing assistance requests.

Name in original language

Camilla - Assistente digitale per i concorsi nella PA

Initiative overview

Camilla is a generative-AI virtual assistant promoted by Formez PA in close collaboration with CSI Piemonte. Formez PA, as the in-house agency of the Department for Public Administration, leads the initiative on the policy and service side, while CSI Piemonte provides and operates the underlying AI platform and conversational technology.

Camilla addresses key challenges in Italy's public recruitment landscape, where digitalisation has increased competition volume and complexity while shortening timelines, often overwhelming candidates with fragmented information. By leveraging generative AI, the initiative taps the opportunity to provide immediate, intuitive access to up-to-date details on competitions managed by RIPAM and Formez PA and hosted on the inPA Portal, Italy's unified gateway for public sector recruitment. This responds directly to user difficulties in navigating notices, schedules, and applications, reducing barriers to merit-based participation.

Formez PA, as RIPAM's operational arm, a role recently enhanced under the 2025 Decreto PA framework, launched Camilla to streamline these processes. The AI assistant draws from a real-time knowledge base via inPA APIs, offering 24/7 text and voice support for queries on open competitions, eligibility, and submissions. A monitored back office refines prompts and content for accuracy, ensuring reliable guidance. Core objectives include enhancing user navigation, cutting staff workload on support tickets, and elevating inPA's digital service quality, aiming for higher satisfaction, fewer inquiries, and broader adoption.

Future evolution focuses on scalability and institutionalisation. The knowledge base will expand from central and regional competitions to encompass all inPA publishers, including agencies, municipalities, and provinces, supported by enhanced APIs, automated verification, and capacity-building for administrations. With monitoring and reporting in place, Camilla is positioned to become a permanent national tool, handling thousands of parallel competitions and evolving via user feedback into a cornerstone of efficient, transparent public recruitment.

Other relevant details

Results, outcomes and impacts: Since its launch in November 2024, Camilla has handled more than 110,000 questions across 33,000 conversations, as of November 2025, with monthly averages exceeding 12,000 queries. This has reduced staff workload on informational tickets and saved users approximately 1,000 hours per year in waiting time for rankings and notices. Results are measured via conversation logs, API analytics, user feedback, and reductions in ticket volumes. Outcomes include higher satisfaction, faster access, seconds rather than hours, and process efficiency. Future plans include full inPA integration by June 2026, scaling to more than 1,000 competitions, with a significant reduction in tickets projected through the expanded knowledge base and monitoring. Challenges: Key challenges included limited user familiarity with generative AI among Italy's population and risks of inaccurate responses due to real-time generation, especially with expanding competition data. These were mitigated via a dedicated back office for 24/7 monitoring, daily and periodic reporting, controlled notice uploads, and iterative prompt and content optimisation in collaboration with CSI Piemonte, ensuring compliance with emerging norms such as the EU AI Act. Lessons learned: Key lessons emphasise structured governance involving Formez PA, the Department of Public Administration, and technical partners; gradual knowledge-base scaling with robust APIs; and ex post feedback loops to evolve the model like a “child that learns quickly”. Success factors include scalable infrastructure, anonymous logging for quality control, and multi-stakeholder coordination, enabling Camilla's growth from 22,000 early queries to more than 110,000.

About the policy initiative


Category:

  • AI policy initiatives, programmes and projects

Initiative type:

  • AI use cases/projects in the public sector

Status:

  • Active

Start Year:

  • 2024

OECD AI Principles:


Images: