AI Agents Conference 2026 - Day 5
summary
A fast, practical batch of war stories and demos: a no-code path uses Codex superpowers to scaffold a single‑page “agent card” app in minutes (author reports a working UI and 50‑token prompt savings); an XGBoost pipeline becomes a multi‑agent HPO that runs ~247 configs automatically and lifts F1 from ~0.84 to ~0.91 (≈7% uplift) while cutting weeks of manual tuning to hours; supply‑chain rerouting agents work on a laptop but expose hard production gaps; anti‑distillation experiments recover ~95% quality from 50% compressed reasoning; CLI/skill patterns shrink projects from ~10k to ~500 LOC.
AI-written summary · about
chapters (14)
- 0:00Start
- 1:20PM (30 min) – Kevin Brown
- 1:55PM (30 min) – Conference Closing & Thank You – Mahesh Chand
- 7:00AM (25 min) – Opening Note – Mahesh Chand
- 7:30AM (30 min) – From Idea to App in Minutes: Building with Codex Skills (No Coding Required) – Rohit Gupta
- 8:05AM (30 min) – From Human-Driven Tuning to Agentic AI: Automating Hyperparameter Optimization in XGBoost – Chaitanya Teegala
- 8:40AM (30 min) – The Hard Truth About AI Agents in Production – Rajesh Gupta
- 9:15AM (30 min) – From Prompts to Agents: Building Intelligent Systems with Foundry Agent Service – Saravanan Ganesan
- 9:50AM (30 min) – Anti-distillation Arms Race in the Age of AI Agents – Oleg Smirnov
- 10:25AM (30 min) – The Best AI Strategy: Don’t Build AI—Build Software Ready for AI Consumption – Pratik Mehta
- 11:00AM (30 min) – A Practical Guide to Evaluation Driven Development – Pratyusha Duvvuri
- 11:35AM (30 min) – AI-Driven Business Transformation – Himesh Bhise
- 12:10PM (30 min) – Metacognition in Agentic Software Development (Agentic SmartDB) – John Godel
- 12:45PM (30 min) – Agentic AI: A New Model for Secure Data Access – Denzil Wessels
related talks
45mAI agents in practice - beyond the prototype
40mHow To Tame Your Agents?
24mAI Agent Swarms Are Amazing
49mAI Agents: Collaboration in Real-Life Applications
40mAI Agents in Practice
34mAI Agents: Transforming Autonomous Decision-Making
40mBuilding Effective AI Agents Simple, Composable Patterns
30mAI Agents Unchained
