OpenAI acquires Rockset to enhance ChatGPT real-time data processing

The ability to index, query, and analyze large amounts of data from real-time sources remains a major headache for anyone involved in building database systems.

Data retrieval takes time and even that assumes that the nature of the queries lie within certain parameters. Overcoming the issue is possible but can be resource-intensive and involve complex engineering. 

Meanwhile, the use cases for real-time data continue to multiply, not least in artificial intelligence (AI) applications connected to cybersecurity automation, fraud detection in financial services, and business analytics in sectors such as manufacturing.

These applications increasingly assume the ability to understand what is happening in data based on a view of the world as it is in the present, not as it was hours or days ago.

Bringing real-time retrieval to AI

These issues go some way to explaining why the world’s most famous AI company, OpenAI, announced last week that it is acquiring the small but interesting real-time analytics database company Rockset in a deal whose terms have not been disclosed.

Founded in 2016, Rockset specializes in solving the challenge of ingesting data in real time. As OpenAI described it in a tweet explaining the acquisition on X:

We’ll integrate Rockset’s technology across our products, empowering companies to transform their data into actionable intelligence.

In a separate announcement, Rockset CEO and co-founder, Venkat Venkataramani, referenced a similar theme:

Rockset will become part of OpenAI and power the retrieval infrastructure backing OpenAI’s product suite. We’ll be helping OpenAI solve the hard database problems that AI apps face at massive scale.

Rockset said that current customers would experience no immediate change in their service but would be moved to OpenAI at some point.

Improving the AI stack

Its first significant acquisition, OpenAI has a lot to gain from the technology Rockset will bring to its platform.

At the moment, this primarily relates to ChatGPT Enterprise, which is focused on using pre-trained AI data models with some ability to train on custom data sets for applications such as business chatbots.

However, its ability to handle real-time data streams is limited beyond some integration with Bing, which hinders its appeal in many contexts.

While Rockset’s technology isn’t AI-specific it is incredibly useful for applications where ingesting real world data is important, especially when using techniques that speed up output such as retrieval-augmented generation (RAG).

Adding Rockset’s real-time capability could also be important as the company tries to scale its enterprise platform longer term to compete with the services being developed by “full stack” AI rivals such as Google and Amazon as well as other chatbots powered by large language models (LLMs).

On the latter front, In March Amazon made a $4 Billion investment in startup Anthropic, which only days ago released the latest version of its highly regarded chatbot, Claude 3.5 Sonnet.

Separately, OpenAI is also reportedly acquiring Multi (previously Remotion), a tiny Mac-oriented collaboration and screensharing company based in New York City.

Multi said it would be closing down after July 24, “after which we’ll delete all user data.” The company’s employees would then join OpenAI.