If you want to know where enterprise data and AI strategy is heading for the next 12 months, the Databricks Data + AI Summit is where that conversation happens. The 2026 edition runs June 15-18 at the Moscone Center in San Francisco, with more than 30,000 data and AI professionals expected in person and tens of thousands more joining virtually.

This year’s theme — “Build Apps and Agents That Work” — is a deliberately honest statement about where the industry actually stands. After two years of agent hype, the practitioner community is asking a harder question: how do you get these things to run reliably in production?

What the Theme Is Really Saying

Most AI agents demonstrated in 2024 and 2025 worked well in controlled environments and fell apart in the real world. Data pipelines broke. Context windows filled up mid-task. Models hallucinated at critical decision points. Governance frameworks weren’t in place when the audit came.

Databricks chose “Build Apps and Agents That Work” as its conference tagline — and that word “work” is doing a lot of heavy lifting. It’s an acknowledgment that agents have not yet delivered on their production promise for most enterprises, and that closing the gap between demo and deployment is the defining challenge for data and AI teams right now.

The co-founders — Ali Ghodsi, Matei Zaharia, Arsalan Tavakoli-Shiraji, and Reynold Xin — are taking the main stage with product announcements, live demos, and customer stories. With 800-plus sessions across tracks covering AI agents, data engineering, governance, analytics, and data warehousing, the event is less a celebration of where AI is and more a working session on where it needs to get to.

Agent Bricks and the Agentic Data Pipeline

The feature drawing the most pre-conference attention is Agent Bricks — Databricks’ architecture for building autonomous agents into the data engineering lifecycle itself. Rather than treating AI agents as a separate layer bolted onto existing workflows, Agent Bricks embeds goal-driven agents directly into data ingestion, transformation, quality management, and orchestration.

The practical implication: a data team using Agent Bricks can describe a data outcome — “keep this table clean, flag anomalies, and notify the business team when something looks wrong” — and the agent handles the implementation continuously. It’s agentic automation applied to the infrastructure layer, not just the application layer.

This sits on top of Databricks’ existing governance stack, including Unity Catalog (for data access control and lineage), Delta Lake (for reliable storage), and Mosaic AI (for model training and serving). The governance layer matters because any enterprise thinking seriously about AI agents needs to know where their data goes, who touched it, and what the agent did with it.

What the Summit Will Tell Us About Enterprise AI in the Second Half of 2026

Several themes are worth watching across the four days:

Evaluation becomes standard practice. The summit has expanded its agentic AI track to include full labs on agent evaluation — how you measure whether an agent is doing what you intended across thousands of runs, not just in a demo. Expect new tooling announcements here.

Data governance is no longer optional. With the EU AI Act’s high-risk compliance deadline approaching in August 2026, and the Colorado AI replacement law (SB26-189) establishing transparency requirements, the governance sessions will be unusually well-attended by legal and compliance teams — not just engineers.

Fine-tuning returns as a serious enterprise strategy. The summit includes focused sessions on fine-tuning models with internal company data — a sign that enterprises are moving past generic foundation models toward purpose-built systems that understand their specific domain, terminology, and process.

What This Means for Business

The Databricks summit is not a news story in the traditional sense — it is a signal about industry direction. When 30,000 practitioners vote with their feet and their conference budget, the themes they pay attention to are the themes that will shape enterprise AI investment decisions for the rest of the year.

Three takeaways for business leaders ahead of the summit:

Data quality is the AI agent bottleneck. Most agent failures in production are not model failures — they are data failures. Agents work from the data they can reach. If that data is inconsistent, poorly labeled, or siloed, the agent reflects those problems at scale and speed. The Databricks focus on data engineering as the agentic foundation is the correct one.

Governance is becoming a competitive differentiator. Companies that invested in data catalogs, access controls, and audit trails two years ago are finding that those investments now translate directly into AI agent readiness. Companies that skipped that step are discovering it cannot be shortcut when regulators start asking questions.

Upskilling the data team is still the critical variable. Every tool on the conference floor — Agent Bricks, Unity Catalog, Mosaic AI, the whole stack — requires people who know how to use it. The difference between organizations that will benefit from agentic data engineering and those that won’t comes down to whether their teams understand data at the level these tools require.

Enterprise DNA has trained more than 220,000 data professionals across Power BI, Python, SQL, and AI tooling for exactly this moment. If your team is heading to the summit, make sure they have the foundations to make sense of what they see there. If they’re not going, the practical upskilling that maps to what the summit will teach is already available through our platform.

The conference agenda opens June 15. The real work starts when practitioners get back to their desks.