The Databricks EMEA Learning Festival

Reliable data is the foundation of every production workload. In this keynote, we unpack the biggest DAIS 2026 announcements — Genie ZeroOps, Genie Code in Lakeflow, Lakeflow Designer, Lakeflow Connect, Zerobus Ingest, SDP Real-Time Mode, and Lakeflow Jobs External Orchestration — and show how they come together in a production-ready data pipeline.

A live demo walks through the stack you'll build in training: Lakeflow Connect, Genie Code, Delta tables, Spark Declarative Pipelines, Unity Catalog, data quality monitoring, and Lakeflow Jobs.

You'll leave with a clear blueprint: build incremental, reusable pipelines; govern data and lineage from ingestion onward; and create a reliable foundation for AI, BI, and operational workloads.
 

TBC Bank, one of the largest banking groups in the Caucasus, faced month-long analyses and 14-week credit risk model deployments across a fragmented on-premises estate.
The bank built a governed Lakehouse on Databricks and adopted Lakebase as the operational layer connecting trusted data with transactional applications. Their DevOps team deployed Lakebase in hours, using its Postgres compatibility to power reverse ETL, AI-driven risk scoring and AML investigation through Databricks Apps.

The results: analyses in minutes, credit risk model deployment in two days, including model risk validation, and more than 600 users querying governed data through Genie. Hear from TBC Bank’s Head of Data Engineering about building a production-ready foundation for AI-driven banking.

Models are a commodity, the agent around them is your product. In this practitioner-led session, we unpack the most impactful announcements from DAIS 2026 — Agent Bricks, Multi-Agent Supervisor with MCP, Unity AI Gateway, and Omnigent — and show how they combine into a governed, production-grade GenAI agent.

A live demo walks through the exact stack you'll build in the hands-on training: AI Search, agent-as-code RAG, MLflow tracing & evaluation, Unity Catalog registration, and deployment with the Review App. 

You'll leave with a clear blueprint: start from the question (not the model), trace and evaluate from day one, and govern early so your agent is production-ready, not just demo-ready.

TBC Bank, one of the largest banking groups in the Caucasus, faced month-long analyses and 14-week credit risk model deployments across a fragmented on-premises estate.
The bank built a governed Lakehouse on Databricks and adopted Lakebase as the operational layer connecting trusted data with transactional applications. Their DevOps team deployed Lakebase in hours, using its Postgres compatibility to power reverse ETL, AI-driven risk scoring and AML investigation through Databricks Apps.

The results: analyses in minutes, credit risk model deployment in two days, including model risk validation, and more than 600 users querying governed data through Genie. Hear from TBC Bank’s Head of Data Engineering about building a production-ready foundation for AI-driven banking.
 

AI has evolved from bespoke models and coarse data boundaries to dynamic, context-aware applications that can retrieve information and take action. This keynote follows that journey through the latest capabilities in Unity Catalog and Lakebase, including vector and text search, fine-grained governance, and governed operational state.

The story moves from analytical data to AI applications and back again: Reverse ETL and LTAP bring trusted silver and gold data into Lakebase-backed experiences, while agent actions become signals in the lakehouse. Together, Unity Catalog and Lakebase point toward AI applications that are trusted, connected, and ready for production.

TBC Bank, one of the largest banking groups in the Caucasus, faced month-long analyses and 14-week credit risk model deployments across a fragmented on-premises estate.
The bank built a governed Lakehouse on Databricks and adopted Lakebase as the operational layer connecting trusted data with transactional applications. Their DevOps team deployed Lakebase in hours, using its Postgres compatibility to power reverse ETL, AI-driven risk scoring and AML investigation through Databricks Apps.

The results: analyses in minutes, credit risk model deployment in two days, including model risk validation, and more than 600 users querying governed data through Genie. Hear from TBC Bank’s Head of Data Engineering about building a production-ready foundation for AI-driven banking.
 

Writing the query is no longer the hard part. Genie will draft it for you, and your colleagues will start asking questions without you. What is scarce now sits underneath the answer: the definitions, the ontology, the curation and the governance that decide whether the answer is right.

This session opens the Analytics track. We start with what was announced at Data + AI Summit - Genie One, Genie Agents, Genie Ontology, Genie Code and AI/BI Dashboards - and what each one changes for an analyst. Then we go under the hood, where you will watch how just a few questions can accelerate what would be weeks of chasing cross-functional teams for an action plan.

You will leave with habits you can start in the Day 1 lab, and a clear reason to finish the Day 2 training on building reliable Genie Agents.

Mercedes-Benz Korea needed every business user — from Sales VP to CFO — to ask questions in plain language and get consistent, trusted answers. But their KPI semantics were scattered across BI reports, so the same question could return different results depending on which report handled it.

The team built an AI-ready semantic layer on Databricks and put Genie at the centre, so every persona gets a governed "Talk to Data" experience. Their key principle: conversational AI doesn't automatically mean trusted AI; trust has to be engineered underneath it.

In this session, we'll walk you through the 5-phase playbook Mercedes-Benz Korea now scales to markets worldwide.