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.

The Interministerial AI Assistant


Added by:   OECD analyst
Added on:   20 Jul 2026
Updated by:   OECD analyst
Updated on:   20 Jul 2026

The AI Assistant is a secure chat tool for public servants, designed to help them write, summarize and find information faster. It was developed to give government teams an easy and trusted way to use generative AI in their daily work. The goal is to improve efficiency while reducing duplicated efforts across ministries.

Name in original language

L'Assistant IA interministériel

Initiative overview

The AI Assistant is a secure, government-wide chat tool designed for public servants. It helps users draft and improve text, summarize long documents, and answer day-to-day questions more quickly, while keeping a clear focus on safe and responsible use in a public administration context. The initiative was developed to address a growing need across ministries: many teams were exploring generative AI, but often with separate tools and duplicated technical efforts. By providing a shared assistant, the government aims to make AI easier to access for public servants, reduce fragmentation, and support productivity gains in routine tasks.

In 2025, the AI Assistant was piloted across 8 ministries with up to 10,000 volunteer users. It helped public servants draft, summarize and search information faster, supporting everyday work and reducing the need for separate ministry-by-ministry tools. Impact has been tracked through user adoption (registrations and active use), usage statistics (queries, features used), and qualitative feedback from pilot administrations. Future results are expected to include wider adoption, more time saved on routine tasks, and improved consistency and security in how generative AI is used across government.

Key challenges included ensuring strong security and compliance for government use, scaling computing capacity cost-effectively, and building trust among users. Another risk was fragmentation, with ministries developing separate assistants and duplicating effort. These challenges were addressed by hosting the service on secure infrastructure, monitoring usage to manage costs, and designing the Assistant as a shared tool that can evolve with ministry needs. Lessons learned include the importance of starting with clear, high-value everyday tasks (writing, summarising), providing user guidance and support, and collecting feedback early to improve the service. Success depends on reliable performance, strong governance, and gradual scaling based on real adoption.

Over time, the AI Assistant is expected to expand its user base and offer more advanced features, including stronger connections to trusted public information sources and the ability to work with internal document collections. It is designed to scale progressively, in line with security requirements and real usage needs, while remaining interoperable with other government digital tools.