What we're looking for

Sessions across the full agentic AI landscape, including (but not limited to):

• Building AI agents – single-agent and multi-agent systems; frameworks like LangGraph, CrewAI, AutoGen, Google ADK, and the OpenAI Agents SDK
• Tool use, function calling, and the Model Context Protocol (MCP)
• Agentic RAG, agent memory, and context engineering
• Orchestration, planning, and reasoning patterns (ReAct, plan-and-execute, reflection)
• Evaluations, observability, and guardrails for agents in production
• Agentic coding and developer workflows
• Real-world case studies - what you built, what worked, what didn't
• Voice agents, automation, and applied agents in industry
• Governance, safety, and the economics of running agents at scale
• Research directions and where agentic AI is heading

Who should submit

Anyone. You don't need to be a well-known speaker or hold a PhD. If you're a practitioner, founder, student, or researcher with something real to share, this stage is for you. First-time speakers are especially welcome - reach out and we'll help you shape your session.

Session formats

• System Prompt (Keynote, 25–30 min) - one talk that anchors the evening and frames where agentic AI is heading.
• Tool Calls (Lightning talk, 5–15 min) - two short, focused talks that go deep on one idea, tool, or lesson. High signal, no fluff.

Levels

All levels welcome, from introductory ("what even is an agent?") to deep technical dives. Please indicate the expected level and any prerequisites in your submission.

Target audience

Software and ML engineers, data scientists, founders, product builders, students, and the broadly AI-curious.