Voltron Data Inc., a well-funded data management startup, today announced that it has acquired Claypot AI Inc. to enhance its product portfolio.

The terms of the deal were not disclosed. Claypot AI previously raised an undisclosed amount of funding from Quiet Capital and Lightspeed Venture Partners. 

“Together we’re going to be able to accelerate our real-time and MLOps product roadmap with state-of-the-art features for our customers,” said Voltron Data co-founder and Chief Executive Josh Patterson.

Voltron Capital launched in 2021 and raised $110 million the following year from a group of prominent investors. The company introduced its flagship product, Theseus, in December. It’s a distributed query engine that companies can use to prepare business data for processing by artificial intelligence models.

The information that an enterprise wishes to input into an AI model is often scattered across multiple internal systems. In such a scenario, the information must be moved into a centralized environment using so-called ETL or extract/transform/load software. In some cases, the records must also be reformatted before processing may begin, which adds further complexity to AI projects.

Once all the necessary information has been prepared, developers have to perform so-called feature engineering. That’s the task of determining which specific subset of the available data points an AI model should consider when making decisions. For example, a neural network tasked with predicting quarterly revenues might evaluate two data points to create a forecast: the number and average price of customer purchases made in the previous quarter.

Companies historically prepared the data they use in their AI projects with central processing units. Theseus makes it possible to perform the task using graphics processing units. The software maker says that its approach can help companies build and run their AI applications more efficiently.

Voltron Data claims Theseus can perform some data processing tasks 45 times faster than Apache Spark. Moreover, the company is promising to reduce the associated costs by a factor of 20. It allows customers to analyze their data without having to move it to a new processing environment, which can further simplify AI initiatives.

The query engine is mainly geared toward batch processing use cases. Batch processing is a term for data projects in which the results of an analysis are not available immediately, but rather take a few hours or longer to arrive. The Claypot AI deal that Voltron Data announced today will extend its market reach to the real-time analytics segment, where the emphasis is on returning data processing results within a fraction of a second.

Claypoint AI was founded in 2021 by Chip Huyen and Zhenzhong Xu, who previously held AI engineering roles at Nvidia Corp. and Netflix Inc., respectively. The startup has developed a product it describes as a “data abstraction” for AI projects. The software can perform both batch and real-time data processing, as well as switch between the two modes when a software project’s requirements change.

Voltron Data will use Claypoint AI’s technology to add GPU-powered real-time analytics capabilities to its product portfolio. The company envisions customers applying the capabilities to use cases such as feature engineering and MLOps. 

Image: Voltron Data