Databricks Data + AI Summit Key Takeaways from Day 2

After announcing a number of important announcements at Databricks’ Data + AI Summit in San Francisco on Tuesday, the company added more context and messaging to introduce its latest products and services during a three-hour keynote session on Wednesday.

Here are five key takeaways from day two.

1. Databricks believes the cost of AI is a key issue for enterprises.

Throughout both days of the summit, Databricks executives repeatedly mentioned the rising cost of AI as organizations seek to contain the costs of token usage and agent development. In an analyst briefing this week, CEO Ali Ghodshi pointed out Databricks’ own cost woes, explaining that agent AI is increasing the consumption of the company’s information technology resources.

Godoshi returned to this theme in his second day keynote today. “This is going to be very expensive,” Godi said. “We’ve just scratched the surface.”

2. Omnigent emphasizes the need for an agent control platform.

The just-released metaharness Omnigent is Databricks’ statement on the need for agent monitoring. Today, the company explained the importance of combining, sharing, and controlling agents to help enterprises manage the increasingly important software that surrounds the language models that interface agents with the world. This is an Omnigent feature.

“This is a harness within a harness,” Databricks co-founder and chief technology officer Matej Zaharia said at the summit meeting. “We need a new open layer above the harness level.”

3. Developers can play a central role in AI governance.

Agent Bricks, combined with recent enhancements to Unity AI Gateway, demonstrated Databricks’ belief that developers play a critical role in AI governance. At the same time, the company wants developers to focus on building agents rather than infrastructure.

Enhancements to Unity AI Gateway provide situational security policies, the ability to monitor costs and set budgets, and smart routing capabilities to find the best model to perform a task. “Agents are the most privileged subjects,” said Kasey Uhlenhuth, Product Director at Databricks. “This is what AI governance in the enterprise should look like.”

4. Databricks is seizing the machine learning market opportunity.

In a presentation on Wednesday, Databricks Product Management Director Mike Del Barso provided a compelling case for how machine learning is an often overlooked but critically important enterprise resource. He also revealed that Databricks is well-positioned to lead machine learning efforts.

The company added more details about AI Runtime, a serverless graphics processing unit for enterprise deep learning that has been in preview for the past two months. Databricks also announced Genie Code for ML and Genie Zero Ops for ML, agents for running machine learning systems.

“What we really want to do is automate ML,” Del Balso says. “We are entering an era where we can scale ML across all businesses. No one has seen ML in the real world more than Databricks.”

5. The world of marketing is undergoing a major transformation with agent AI.

Working with language models at scale is transforming agents into buyers and reshaping the way marketing and data teams build new initiatives. The announcement of CustomerLake, an agent customer data platform (CDP) by Databricks, provided another example of how the marketing space is being transformed.

CustomerLake is designed to help marketers move from static, one-time campaigns to infinite campaigns, agent-driven engagement loops that analyze customer signals and act across channels.

“People don’t buy things for themselves anymore,” Tasso Argyros, vice president of engineering at Databricks, said today. “Commerce is becoming headless. What if every customer had an agent? We think this is the future.”

Photo: Mark Albertson/SiliconANGLE