Piotr Paczewski

Multiple engines and catalogs make governance hard and AI‑ready data even harder. In this session, we will show how to use Snowflake Horizon and Snowflake CoCo to build an interoperable lakehouse that is secure, well-governed and easy to extend. We will start by designing Iceberg-based lakehouse, then use Horizon to centralize discovery and policy enforcement across engines. On top of this governed foundation, we will introduce Snowflake CoCo as the developer copilot that generates and refactors the data lake objects. Attendees will leave with patterns and code templates to operationalize an interoperable data lake on Snowflake.

Discover how Snowflake CoCo skills can transform complex geospatial deployments into guided, repeatable workflows. This session demonstrates a real-world Fleet Intelligence solutions for retail, manufacturing, and travel and hospitality that perform sophisticated geospatial analytics, all deployed through an AI-assisted skill. Attendees will learn how to package domain expertise into reusable skills that handle template variables, validate prerequisites, execute idempotent SQL and provide self-healing error recovery, making advanced geospatial solutions accessible to any Snowflake user.

This session demonstrates how to transform natural language requests into deployment-ready geospatial applications using Snowflake CoCo. You will learn how to parameterize solution templates and extend logic to support organization-specific KPIs. Discover how Snowflake enables businesses to optimize delivery routes, increase fleet utilization, and reduce both idle time and operational costs. Attendees will gain practical knowledge on using Snowflake CoCo to deploy and customize geospatial solutions within the Snowflake native stack. All session materials, including code and templates, will be made publicly available for immediate use.

Join this session to explore the key concept and learn how to build a simulated LLM-powered route optimization app using Streamlit and Carto’s Overture maps for delivery planning. The app leverages Cortex LLMs to filter maps, generate customer samples and optimize routes. The process is documented in a Lab notebook, resulting in a dynamic visualization within Streamlit in Snowflake. The backend integrates SPCS and a Docker image of an open-source route optimization service to generate routes from real data. The app will be published as a Snowflake Native App.