Databricks expands its vision for data-driven enterprise

Databricks is on a roll today as it unveiled a new offering for structured data, a better approach to building AI agents, and more.

Over 20,000 Databricks fans, I mean users, were in San Francisco this week, attending what's among the top tech conferences around.

New announcements

Here are a couple of announcements that caught my attention.

a. Lakebase

Lakebase is a managed database that's now available on Databrick's Data Intelligence Platform.

• Uses Postgres for structured data.
• Novel separation of compute, data.
• Compute is powered by serverless.
• Design yields single-digit latency for OLTP.

b. Agent Bricks

Doing AI agents is not as easy as it looks, says Databricks CEO Ali Ghodsi. But Agent Bricks will make it much easier to create AI agents.

The focus here? The built-in ability to evaluate performance and optimise agents instead of relying on "gut checks."

Better over time

What will the future bring for AI?

While Databrick's chief AI scientist Jonathan Frankle demurred from giving a direct response, he suggested that the history of past inventions could offer useful clues.

• The Internet.
• Web browsers.
• Generative AI.

You see, while the "zero to one" moment always gets the most attention, the technology never stopped getting better.

Think about it: the Internet today is completely different from the Internet of a decade ago. Ditto to Web browsers, which went from static HTML to supporting full-blown apps like design software (Figma) and productivity suites (Google Docs) today.

If there’s one takeaway, it’s this: the AI of tomorrow will likely look very different from what we have today.

It's the data, silly

Of course, surely everyone at a data and AI conference would say data is important. To me, the proof is in the pudding.

Yesterday I wrote about how Standard Chartered Bank used Databricks - a unified analytics platform - to enhance its SIEM. And it makes sense, says Lavy Stokhamer, because cybersecurity has always been inextricably tied to data.

Today, I heard from Santhosh Mahendiran, chief data and analytics officer of Techcombank in Vietnam. Within just 2.5 years, he built the culture, know-how, and tech stack to leverage data to personalise at scale - to a segment of one!

I'll write more another day - it's jaw-dropping.

In conclusion

The quote that stuck in my mind?

What JP Morgan's CEO Jamie Dimon said in a fireside chat: "There's a lot of data out there... [the hard part is] getting the data in the form where it's usable."

What do you think?