Agentic AI: Building Agents We Can Trust
Panel 1 · Cross-AI Symposium 2026
About the Panel
Agentic AI systems no longer just answer questions. They plan, call tools, write code, query enterprise data, trade, and run infrastructure on their own. When an agent acts and something goes wrong, can we trace what it did, explain why, and stop it in time? The panel treats accountability as a technical problem: how to build agents whose actions are reliable, verifiable, auditable, and safe, even as they learn and evolve.
Connected Sessions
Perspectives Sought
- Agent architectures and multi-agent systems
- Tool use and planning
- AI safety, alignment, and self-evolving systems
- Verification and formal methods
- Evaluation, monitoring, and observability of agents
- Reliable data access and NL2SQL
- Agents for finance and big data
How to Apply
To apply as a panelist or moderator, follow the instructions in the Call for Panelists, which covers application requirements, where to apply, deadlines, and what happens after selection. Applications are due November 16, 2026 (AoE).