In this moment, excerpted from the “Who’s Afraid of Little Old Copilot” session, Microsoft’s Dona Sarkar delivered a keynote that captured the industry’s accelerating transition from generative AI experimentation to the emerging era of agentic AI.
Key Takeaways
- From Generative to Agentic AI: In only two years, the industry has already moved past generative AI experimentation into agentic AI, shifting the conversation from producing answers to solving problems.
- Self-Correcting Systems: Using her memorable analogy — “AI that has gone to therapy, unlike your ex” — Sarkar explained that agentic AI can recognize its own mistakes and fix them in real time rather than simply delivering a static response.
- Agents Work Together: Agentic AI is not a single model but a collection of collaborating sub-agents. One generates answers while others evaluate, validate, and refine outputs behind the scenes to improve accuracy.
