The State Department is exploring the implementation of artificial intelligence (AI) agents capable of taking actions on behalf of employees, as stated by the department’s Chief Information Officer, Kelly Fletcher, during an event on Wednesday. The department already features a generative AI chatbot named StateChat, which was launched in the previous year. This chatbot is designed to assist with tasks such as translations and answering inquiries from the department’s foreign affairs manual.
Fletcher emphasized that the goal is to develop AI agents that can perform tasks for users, saying, “I want it to not only tell me, ‘How much leave do I have’ … but then I want it to put in my leave slip, which is in a different system. We’re building to that.” This capability distinguishes AI agents from traditional generative AI by enabling them to take definitive actions.
The AI company Anthropic, known for its language model Claude, is also focused on the development of action-taking AI. Co-founder Jack Clark mentioned at a forum in Washington D.C. that by late 2026 or early 2027, the company aims to create systems that can autonomously complete complex tasks that typically require extensive time.
While agentic AI presents opportunities for automating operations and boosting productivity, it brings risks such as oversight challenges, evaluation difficulties, and the potential for job displacement. Greg Barbaccia, the government’s chief information officer, has expressed the intent to leverage AI to compensate for workforce reductions in the federal government, notably after the Trump administration lost thousands of employees, including the State Department’s former Chief Data Officer and AI Officer, Matthew Graviss, who departed in February.
Fletcher acknowledged that the adoption of AI at the State Department has been slow. The initial rollout of StateChat involved 3,000 beta testers, but it has now expanded to be used by approximately 45,000 to 50,000 of the department’s 80,000 employees. She noted that extensive education and training were required to facilitate this user engagement.
She recounted a recent experience where she addressed a question about the chatbot’s usage, highlighting that significant training and conversation efforts were necessary to encourage its adoption among potential users. The chatbot assists State employees by guiding them through internal policies, offering scenarios such as how to relocate a pet cat.
Fletcher’s vision includes consolidating the department’s administrative functions behind a single chatbot and integrating additional agents focused on specific mission areas. Looking ahead, she believes AI will be embedded across various operations, including prioritizing cybersecurity alerts, and the department is also piloting a chatbot for its electronic health record patient portal.
Fletcher concluded by emphasizing the importance of smart integration and user awareness regarding the applications of AI technology within the department.
