Source summary On September 16, OECD published a report on AI agents in organizations. Interviews with 25 organizations—including developers, adopting businesses and public bodies—describe the move from experimentation to operational use and differences in organizational maturity. Approaches include gradually increasing autonomy while retaining human checks for consequential actions. Evaluation and assurance, cybersecurity, accountability and traceability remain challenges. TABS Company's business perspective Start with clearly bounded work such as document classification or drafting internal reports, and define success criteria. Require staff approval for consequential steps such as external communications, contracts and payments. Design access permissions, execution logs, stop conditions and recovery procedures together. Measuring accuracy and review time alongside automation rates can support decisions about expansion. Interpretation This is qualitative research on early practitioner experience. It should not be treated as a representative survey of all businesses or as evidence of a guaranteed investment return. This is an unofficial English translation of TABS Company's Korean summary and adaptation of an OECD work. The opinions and arguments expressed do not necessarily reflect the official views of the OECD or its member countries. In the event of any discrepancy, the original OECD work prevails.