Streaming vendor Redpanda buys SQL engine, unveils AI suite

Addition through acquisition

SQL engines are a hot commodity.

The Oxla acquisition appears to be a good fit and delivers Redpanda with a critical missing piece around standard SQL capabilities that agents need to query both streaming and at-rest data. This acquisition allows Redpanda to offer a complete data access layer rather than just streaming capabilities.

SQL databases date back to the 1970s and are relational databases that use structured query language (SQL) to manage data in columns and rows. These databases are currently having a renaissance because they can make it easier for developers to build and manage AI tools, including agents.

AI models require large volumes of high-quality, consistent data to be accurate. One of the strengths of SQL databases is that they maintain data integrity and consistency. In addition, SQL databases easily enable complex joins so users can integrate data from different sources, support vector indexing to fuel similarity searches for discovering relevant data, include built-in machine learning functions and provide governance and compliance capabilities.

Last spring, Databricks and Snowflake each made acquisitions to add SQL database capabilities to their AI development suites. Redpanda is following suit with the acquisition of Oxla, which is based in Warsaw, Poland, and raised $11 million in funding before its purchase.

With Redpanda Connect, the vendor enabled customers to process data -- in particular, unstructured data -- to train agents and applications. A SQL engine will allow Redpanda users to operationalize structured data, according to Tyler Akidau, the streaming specialist's chief technology officer.

"Having a SQL engine opens up this whole world of data transformation and data analysis that is super valuable for agents," he said.

In addition, having a SQL engine will better enable Redpanda to adapt to customers' agentic needs rather than force customers to alter their systems to work with Redpanda's platform, Akidau continued.

"Any one agent isn't going to connect to all of an enterprise's data, but in the aggregate, a large fleet of agents is effectively going to connect to all of an enterprise's data across everything," he said. "Having a nimble SQL engine that brings agents closer to the data spread across the enterprise is really valuable."

Catanzano similarly noted the importance of providing a SQL engine to help enterprises connect their data.

"AI agents need a standardized interface to query diverse data sources in real-time, and SQL provides that universal language," he said. "Agents require seamless access to both historical data for context and real-time streams for current state, making SQL the bridge that enables agents to operate … without requiring specialized query languages for each data source."

Michael Ni, an analyst at Constellation Research, like Catanzano noted the significance of the acquisition given that it helps Redpanda expand beyond its streaming data roots and could help the vendor differentiate itself from competitors, such as Confluent.

"Oxla turns Redpanda from a streaming engine into a full-fledged decision fabric," he said. "It's the move that makes its real-time data accessible to both humans and agents through familiar, governed queries. … Redpanda can [now support] multimodal data access for agentic reasoning, a differentiator none of the pure streaming vendors yet match."

Redpanda could have built its own SQL engine but elected to buy such capabilities because of how difficult it is to develop them, according to Akidau.

"Anyone who has ever tried to build a SQL engine will tell you it's a massive undertaking," he said. "This acquisition gives us years of advantage over building it ourselves."

Managing agents

While the acquisition of Oxla helps developers and agents access data, the ADP is Redpanda's complete environment for training and managing AI applications.

OpenAI's November 2022 launch of ChatGPT marked significant improvement in generative AI (GenAI) technology and sparked a surge of interest in AI development among enterprises. Initially, AI development focused largely on building GenAI chatbots that made workers better informed by enabling them to query and analyze data using natural language rather than code.

In 2024, that evolved to focus on building agents, which can be trained with contextual awareness and reasoning capabilities that enable them to act autonomously so they not only help inform employees, but make business processes more efficient.

To help customers build AI tools trained with proprietary data, many data management vendors have created environments within their platforms specifically aimed at training and managing agents.

For example, Databricks provides Agent Bricks and Snowflake offers Cortex Agents. Confluent, Informatica, MongoDB and Teradata are among others that have introduced agent training capabilities.

The ADP is Redpanda's version.

Consumer AI has been successful whereas most enterprise AI pilots have failed to make it into production, Akidau noted. Conversations with customers struggling to successfully develop AI tools, therefore, provided Redpanda with the impetus for creating the ADP.

"A picture emerged of challenges being around taking AI agents, which are super-human in a lot of their capabilities but also highly chaotic and unpredictable, and letting them loose in private networks on private data," Akidau said. "Following that thread is what led to what we have here."

Redpanda does not provide development tools within the ADP. Instead, Redpanda's new suite is geared toward accessing relevant data -- including streaming data -- and providing governance while integrating with development frameworks.

"Our goal isn't to build the end-all-be-all development framework," Akidau said. "Our value proposition is to work with whatever tools customers use and integrate their agents into this governance layer."

Featuring connectivity through MCP and A2 with built-in observability, querying, and transformation capabilities, the ADP is logically built to provide Redpanda's customers with a suite for training and governing agents, according to Ni.

However, it's only a beginning, he cautioned, and could benefit from a semantic layer.

"Redpanda's Agentic Data Plane brings together many of the right components such as event streaming, distributed SQL, Iceberg support, and governance," he said. "Redpanda's announcement provides the execution core needed to ground agentic AI in real-time data, but it's an early-stage fabric that needs semantic modeling and workflow integration to be enterprise-complete."

Looking ahead

Redpanda's primary product development focus for the remainder of 2025 and into 2026 is to make all aspects of the ADP generally available, according to Akidau.

Focusing on the ADP is wise, according to Ni. However, rather than merely make the capabilities that currently comprise the vendor's new suite generally available, he suggested that Redpanda add a control pane where agentic training and governance are unified.

"To win the next stage, Redpanda must fill out and mature its stack, extending its governed data plane to an agent control tower where real-time context meets intent, and every enterprise decision can not only be served with real-time data but observed, explained and optimized," Ni said.

Catanzano, meanwhile, advised that Redpanda expand the ADP by adding capabilities such as prebuilt development templates and a hub where customers can discover agents built by other users and optimized for the streaming data specialist's platform.

"Redpanda could develop prebuilt agent templates and frameworks for common enterprise use cases -- customer service, data analysis or workflow automation -- to lower the barrier to entry for organizations wanting to deploy agentic AI," he said. "Additionally, they could create an agent marketplace … similar to how cloud providers offer application marketplaces."

Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.