Agentic AI White Paper

ATARC’s mission is to bring together academia, industry and government to tackle challenges in IT modernization and emerging technology adoption. The next phase of AI’s evolution is agentic AI – systems that are given goals and the ability to interact with the world around them, and which can formulate plans and execute actions to achieve the given goals. The future of agentic AI is transformative; their ability to utilize tools, take actions, reflect on their performance, and manage both short and long- term memory will usher in a new era of automation. Multi-agent systems are extending the reach of foundation models by integrating external knowledge, mitigating hallucinations, and enhancing security. These advancements are not merely incremental; they represent a fundamental shift in how we interact and leverage artificial intelligence.

This paper provides an overview of Agentic AI, use cases, risk frameworks, governance and best practices. Areas of focus include: establishing and enforcing guardrails around uses of agentic AI, ensuring the security and ethical application of these systems, and protecting the “supply chain” of their hardware, software, and development data.