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.

"Services Publics+" : Using AI to improve the quality of public services and re-humanize public services


Added by:   OECD analyst
Added on:   29 Oct 2026
Updated by:   OECD analyst
Updated on:   29 Oct 2026

France’s "Services Publics+" programme uses AI to improve the quality of public services for millions of citizens. It was created to reduce common frustrations, like long wait times, and help civil servants work more efficiently. By analysing citizen feedback and drafting quicker responses, the programme helps administrators identify problems and deliver a better, simpler experience for everyone.

Name in original language

Services Publics+ : utiliser l’IA pour améliorer la qualité des services publics et les rendre plus humains

Initiative overview

The French Government’s Services Publics+ (SP+) programme is a nationwide initiative designed to restore the fundamentals of public service and improve the daily experiences of 45 million citizens interacting with 2.5 million public servants. Historically, citizens have faced frustrations such as long wait times, complex forms, and a lack of transparency. To resolve these issues and leverage the capabilities of emerging technologies, the Government has integrated generative AI into the SP+ platform—specifically through its citizen feedback feature, "Je Donne Mon Avis" (I Give My Opinion). This transforms how user feedback is collected, analysed, and actioned across major public services, from police stations to social benefit desks. The primary objective is to drive continuous improvement at all contact points, making services more efficient while freeing up civil servants to focus on strategic, high-value tasks.

To achieve this, the initiative deploys three core AI components: speech-to-text capabilities to ensure accessibility for mobile users, individuals with disabilities, and those facing digital illiteracy; automated drafting assistance that provides civil servants with proposed responses to citizen feedback, which they can easily review and refine before sending; and real-time semantic analysis of user feedback across the public services, creating summaries that detect weak signals, pinpoint administrative difficulties, and suggest concrete action plans. These summaries can also be generated for a specific irritant.

These technological integrations are already demonstrating measurable value. The initiative currently yields three main benefits: improving public satisfaction and civic engagement; streamlining service delivery by reducing the time civil servants spend on trivial or repetitive tasks; and reducing response times to make the administration more agile. As the programme matures, the key challenge for the Government is comprehensively assessing and quantifying the long-term impact these AI tools have on the actual quality of service experienced by both users and civil servants.

Looking forward, the initiative is set to scale by deeply institutionalising these AI insights into both local governance and public transparency. The AI-generated feedback summaries will soon be provided directly to local préfets (State representatives), enabling them to closely monitor and quickly resolve specific bottlenecks within their départements. Additionally, these overviews will be made publicly available, giving the general public a transparent, aggregated view of what their fellow citizens think about each public service. This evolution is intended to ensure a continuous, transparent loop of civic feedback and administrative modernisation.

Other relevant details

Results, outcomes and impacts: Impact is measured via four KPIs: response time, AI-agent text similarity, agent feedback, and user satisfaction. Observed results include faster service: response times went from 19 to 3 days; agent efficiency: civil servants validate 70% of AI-drafted text, and 75% find the tool useful; and citizen satisfaction: 68% of users find AI-assisted replies useful, an 11-point increase from before (57%). In the future, impacts are expected to expand as AI summaries are given to local préfets to resolve regional issues and shared publicly for national transparency. Challenges and lessons learned: Tailoring the LLM to the specific needs of all distinct ministries and public services requires continuous adaptation. Furthermore, deploying public-sector AI poses significant risks regarding GDPR compliance, cybersecurity, and the protection of fundamental rights. To support trustworthy AI, all data is anonymised and published as open data. We established strict "human-in-the-loop" guardrails: public agents can opt in or out, must validate or edit AI drafts, and remain solely responsible for final responses. Following the AI Act's transparency principles, citizens are explicitly notified when an answer is AI-assisted. We learned that AI efficiency relies entirely on human expertise; agents are essential for evaluating outputs and escalating issues for LLM re-engineering. Success requires absolute transparency, robust legal compliance, and continuous human agency to maintain public trust.

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:

  • 2023

Target Sectors:


OECD AI Principles:

—

Other relevant urls:

—