Databricks

When data grows faster than the systems around it, complexity becomes the default. That was the challenge for Indra Renewable Technologies, one of the fastest-growing electric vehicle (EV) charging companies in the United Kingdom, which designs, engineers, and manufactures smart chargers for commercial and home locations. As their data estate expanded, so did the number of tools, pipelines, and costs attached to it.

We all know the pattern: you ask an AI tool a question and get an answer in seconds, only to spend the next hour investigating the accuracy, reformatting insights into your reporting template, and shipping your update—all while a slew of other open tabs fight for your attention. The first answer might be instant, but the work that follows is still painfully manual. Since Data + AI Summit, we've been working on what happens after Genie One gives you an answer.

Construction generates abundant data, from equipment telemetry and maintenance records to job-site documents and rental feeds. But like manufacturing, the industry faces fragile supply chains, margin pressure, and rising demands for speed, customization, and traceability. The data remains fragmented across systems, organizations, and equipment types. Ownership can change from project to project, and critical information may be unstructured, disconnected, or never captured.

At scale, your training efficiency is determined by a single metric: "goodput", the proportion of time your GPUs spend on productive computation rather than waiting or recovering from failures. Because GPU failures are the expected case at scale, the ability to rapidly and automatically recover from a failure is the only way to maintain high goodput and manage your total GPU spend.

We are headed to VLDB 2026 to share multiple innovations that power the Databricks platform. Databricks Co-founder and Chief Architect, Reynold Xin, will kick-off the conference with the opening keynote. Databricks has four accepted papers related to Lakebase, Spark Structured Streaming, and automatic Lakehouse optimizations. The demo paper on the Enzyme engine will showcase how we incrementally maintain materialized views. Below is a preview of these presentations.

Limited-time offers can be a powerful way to create excitement, bring guests back, and drive incremental sales, but even a great concept can fall short of expectations. When it does, did the offer miss the mark? Or is traditional data analysis not telling the full story? At first glance, the conclusion may seem obvious: guests did not respond. But what if the offer is exceeding expectations in restaurants where it is available?

The Motivation More and more enterprises are now asking agents to work with their proprietary documents and answer questions about their contents. However, much of the important information lives inside figures and charts. Many customers have been finding that agents struggle to answer questions that require reading and counting values in charts. For agents to work reliably in diverse enterprise settings, we need to make charts more interpretable.

In the first blog of this series, we looked at how Lakebase Postgres is rewriting the foundation of enterprise applications - collapsing the decades-old divide between operational and analytical systems into a single governed platform. By bringing a serverless, Postgres transactional database directly onto the Data and AI Platform, Lakebase eliminates the pipelines and duplicate governance that used to sit between a transaction and a decision.

Agents that interact with a traditional OLTP database often create bottlenecks at the storage layer. New deployments, copies, restores, and replicas all mean moving around large volumes of data which is time-consuming and expensive. The polar opposite is true for object storage. Amazon S3, for example, is cheap, performant, almost invisible to operate. It creates a scalable, cost-effective storage layer for agent memory.

Your FinOps lead needs to easily drill-down into Databricks spend and identify what’s driving costs, without maintaining dozens of queries and dashboards. Your data governance team built their own tools to track classification coverage, and they're already stale. Your AI deployment team needs to understand what models and agents are being used by developers. These are the conversations we heard again and again from customers.