AI Engineer World's Fair 2026 — unofficial import demo

In the last two years, models have gotten exponentially smarter. Two years ago they couldn't pass the bar. Today, top 1% of test scorers. And yet most agents still can't answer a simple business question correctly. You ship a demo that works. You deploy it. The business abandons it in a month.

The missing variable is context: the business definitions, procedural knowledge, and operational norms that make a human expert valuable.

Drawing on hundreds of production deployments, Prukalpa Sankar will break down what it actually takes to give agents contextual intelligence — and get them past the demo stage.

She'll walk through the architecture of a context layer: how context repos work (versioned, testable, portable), how simulation environments catch failures before deployment, how agent traces compound back into shared context, and why context engineering scales where fine-tuning and prompting don't. She'll also cover why your context needs to be open (MCP, Iceberg, deploy to any framework) — and what happens when it isn't.