Publication of the Report on Agentic AI in organisations

The Agentic AI project is a GPAI project that takes stock of the landscape of agentic AI systems. It is conducted jointly by the OECD Secretariat, the Tokyo Centre of the GPAI Expert Community and the OECD.AI expert group established for this project.

The project comprises two main strands of work: taking stock of existing definitions and providing a conceptual foundation for the field, and, building on this, analysing use cases and future perspectives. The first strand was led by the OECD Secretariat and has already been published as " The agentic AI landscape and its conceptual foundations " (OECD Artificial Intelligence Papers, No. 56, February 2026). The Tokyo Centre has been involved in the latter, examining use cases and their implications. To this end, the Centre invited Dr. Merve Hickok, an expert in AI policy and President of the Center for AI and Digital Policy, to take the lead and has conducted interviews with twenty-five organisations across the public and private sectors.

The resulting report has now been compiled and published on the OECD website below. Drawing on those interviews, it analyses how agentic AI is being developed, deployed and governed within organisations.

Among the report's findings:

・Agentic AI is rapidly moving beyond pilot projects and into organisational workflows.

・Organisations are adapting existing governance frameworks and introducing safeguards, such as phased increases in autonomy and human oversight of high-stakes actions.

・Significant challenges remain in evaluation, cybersecurity and accountability.

The report contributes new empirical evidence to inform future research and policy discussions on agentic AI.

Link to the Report

Agentic AI in organisations: Early insights from practitioner interviews

Link to the OECD blog post

Putting agentic AI systems to work