Connect your AI to data: implementing the open infrastructure underneath it
Partner Capabilities: Delivery
November 4, 2026
Session 1: 3:00 PM GMT / 10:00 AM EST Session 2: 4:00 PM PST / 9:00 AM AEST (next day)
Virtual
Angel Hernandez
Martin Knazovicky
The architecture pattern: how Fivetran lands data as Iceberg and Delta Lake tables in the client's own cloud storage, how Snowflake and Databricks query it externally, and how bronze and silver layers in the lake feed the warehouse gold layer without reloading or re-ingesting data.
The "yes, and" delivery motion: how to position open data infrastructure as complementary to existing cloud platform contracts rather than a rip-and-replace, and what that means for how you scope and price the engagement.
Security and governance in practice: private networking, write credentials, and how governance stays with the client's existing compute engine catalogs, so you can deliver this in regulated, compliance-heavy environments.
From data to agent, live: a full end-to-end build, from source systems through Iceberg/Delta Lake tables to a working connection into Claude via dbt MCP, showing exactly where the AI connection happens and how to replicate it.
You're all set, [Name].
Registration closed.
AI agents are only as good as the data layer underneath them. This session is a technical walkthrough of how to build that layer, and connect it. Fivetran lands data as open Iceberg and Delta Lake tables in the client's own cloud account, managing ingestion, schema evolution, deduplication, and table maintenance along the way. That same governed data becomes queryable by Claude, ChatGPT, and the client's AI tooling of choice, with no re-ingestion and no vendor lock-in. The session covers the full architecture, the security model, and the "yes, and" framing that makes this a conversation you can have in any account, regardless of what cloud contract they already have, then ends with a live, end-to-end build: starting from source systems and finishing with a working connection into Claude through dbt MCP.
Who should tune in:
Solution architects and delivery consultants building or scoping data platform engagements who want to deliver an AI-ready data foundation as a concrete deliverable, not just a positioning conversation.
The architecture pattern: how Fivetran lands data as Iceberg and Delta Lake tables in the client's own cloud storage, how Snowflake and Databricks query it externally, and how bronze and silver layers in the lake feed the warehouse gold layer without reloading or re-ingesting data.
The "yes, and" delivery motion: how to position open data infrastructure as complementary to existing cloud platform contracts rather than a rip-and-replace, and what that means for how you scope and price the engagement.
Security and governance in practice: private networking, write credentials, and how governance stays with the client's existing compute engine catalogs, so you can deliver this in regulated, compliance-heavy environments.
From data to agent, live: a full end-to-end build, from source systems through Iceberg/Delta Lake tables to a working connection into Claude via dbt MCP, showing exactly where the AI connection happens and how to replicate it.
Angel Hernandez
Martin Knazovicky
Angel Hernandez
Martin Knazovicky
AI agents are only as good as the data layer underneath them. This session is a technical walkthrough of how to build that layer, and connect it. Fivetran lands data as open Iceberg and Delta Lake tables in the client's own cloud account, managing ingestion, schema evolution, deduplication, and table maintenance along the way. That same governed data becomes queryable by Claude, ChatGPT, and the client's AI tooling of choice, with no re-ingestion and no vendor lock-in. The session covers the full architecture, the security model, and the "yes, and" framing that makes this a conversation you can have in any account, regardless of what cloud contract they already have, then ends with a live, end-to-end build: starting from source systems and finishing with a working connection into Claude through dbt MCP.
Who should tune in:
Solution architects and delivery consultants building or scoping data platform engagements who want to deliver an AI-ready data foundation as a concrete deliverable, not just a positioning conversation.