20 min

From Prompts to Pipelines: The Rise of the Agentic Harness - For the last two years, the developer community has been obsessed with "prompt engineering." But as AI scales into production enterprise systems, the prompt is now only 10% of the work. The remaining 90% is the Harness—the engineering pipelines, tool orchestrations, and execution sandboxes that allow an agent to take multi-step autonomous action. In this luminary talk, Romain explores why the next generation of builders must pivot from "prompting" to "agentX (agent experience) engineering." Drawing on his front-row view of world-class agent deployments, he will provide a high-level architectural overview of the Agentic Harness, touching on the critical shift toward persistent memory, secure execution environments, and automated evaluation loops. Attendees will walk away with a high level blueprint for transitioning from building fragile chatbots to architecting robust, enterprise-grade agentic pipelines.

20 min

Paving the Roads for Agents: Rebuilding Infrastructure for the Autonomous Era - Dropping an autonomous AI agent into a traditional software pipeline is like putting a sports car on a dirt road—it will inevitably break down. Today's enterprise tech stacks were built exclusively for human engineers. So, what happens when your most active code contributors become machines? In this session, Factory CTO & Co-Founder, Eno Reyes explores the fundamental infrastructure shift required to support an autonomous engineering workforce. Moving beyond simple code generation, Reyes will discuss the industry-wide necessity of "paving the roads" for agentic systems. He will illustrate how the entire Software Development Life Cycle (SDLC) must evolve, offering builders a practical framework for transitioning to agent-native environments. Attendees will learn how to establish the deterministic guardrails, AI-readable APIs, and robust data foundations necessary for autonomous systems to build, test, and deploy at full speed.

90 min

Hands-On Lab: Building an End-to-End AI Application on Snowflake: From Data to Intelligence - Build a complete AI retail app entirely in Snowflake. Start by using Cortex Code to generate SQL and dbt models for metrics such as CLV. Next, train a GPU-accelerated XGBoost model in Snowflake Notebooks, register it, and deploy it as a service. To handle OLTP, use Snowflake Postgres and sync data to Apache Iceberg on S3, demonstrating lakehouse interoperability with pg_lake. Finally, unify the stack with Snowflake Intelligence, a conversational interface orchestrating Snowflake Cortex Analyst, Snowflake Cortex Search and your ML models for a seamless analytical experience.