A single agent is not a team.
Agentic AI becomes useful in production when agents are specialized, coordinated, governed, observable, and able to work on real enterprise data with clear oversight.
Building AgenticAI Teams onDatabricks
From Pilot to Production. No theory, no generic intros, and no marketing talk: just real-world engineering patterns for specialized, coordinated, and governed agent systems.
60'
Deep engineering
30'
Live Q&A
Databricks
Experience required
Matthias Baumann
View speaker profile & credentials
Live Webinar
Move agentic AI beyond impressive demos.
Matthias Baumann
Group CTO · Ultra Tendency · Databricks Resident SA
No spam. Unsubscribe anytime. Session is free.
Webinar at a glance
Production-ready agent systems are architecture work.
Databricks now has the full stack.
We will connect Agent Bricks, the Multi-Agent Supervisor, MLflow 3, and Managed MCP into one practical engineering view of how agent teams move from pilot to production.
The failure mode is usually architecture.
In client work, many AI projects fail to reach production not because the model is weak, but because coordination, governance, and operational design were treated too late.
Meet your speaker
Matthias Baumann
Group CTO, Ultra Tendency. Trainer Lead, Ultra Tendency Academy. Databricks Resident Solution Architect.
This webinar is led by Matthias Baumann, who brings practical experience from platform implementations, technical enablement, and architecture work across modern data and AI environments.
The session is built around engineering depth, delivery patterns, and the questions technical teams actually face in practice.
Architecture practitioner
Enablement expert
Real-world delivery focus
"No marketing. No generic intros. Just real-world engineering experience."
What you will learn
Real Databricks agentic engineering, from stack choice to governance.
Use the right Databricks agentic layer
Understand when to use AI Functions, Agent Bricks, and the Multi-Agent Supervisor inside the Databricks agentic stack.
Coordinate specialized agents
See how a supervisor agent decomposes complex requests, delegates to sub-agents, and synthesizes final results.
Improve quality over time
Learn how natural language feedback from domain experts can improve coordination quality without turning governance into an afterthought.
Apply Managed MCP in practice
Work through governed access patterns for Unity Catalog tables, Vector Search, and external tools.
Design for auditability
Build agent teams that are auditable and resumable by default, with oversight embedded in the architecture.
Avoid common architecture mistakes
Identify the three most common mistakes teams make when building their first agent team, and how to avoid them.
Engineering focus
From pilot to production means designing the team, not just the prompt.
What the session avoids
- No introductory AI overview
- No generic agent demos disconnected from data governance
- No marketing walkthrough of product features
What the session covers
Specialization
Coordination
Governed access
Auditability
Target audience
Built for technical teams already working with Databricks.
Data Engineers
For platform builders
Engineers responsible for data products, governed access patterns, and reliable production workflows on Databricks.
ML Engineers
For operational AI practitioners
Practitioners moving from model experimentation toward agentic systems that need evaluation, feedback, and production oversight.
Tech Leads
For architecture decision makers
Leads who need to shape standards for agent teams, supervision, governance, and scalable delivery.
Prerequisite
Databricks experience required
This is not an intro session. The webinar assumes practical familiarity with the Databricks platform and focuses on engineering decisions.
Register now
Join the next Ultra Tendency Academy webinar.
Building Agentic AI Teams on Databricks, 17.06.2026, 16:00-17:30 CEST. Online webinar, 90 minutes: 60 minutes of engineering content plus 30 minutes of Q&A.