OpenAI is acquiring the start-up Rockset. The company operates in enterprise analytics.
OpenAI is acquiring both the technology and Rockset’s entire team, Bloomberg knows. Through an integration, enterprise data should become more usable by business users and developers subscribed to OpenAI’s services. This should be accomplished with Rockset’s technology, which makes data searchable, collects, and indexes it. Data can also be accessed in real-time.
According to OpenAI, the acquisition brings opportunities for developers to build more intelligent applications. “Rockset’s infrastructure enables companies to turn their data into actionable information,” said Brad Lightcap, COO at OpenAI.
Commercializing AI
For OpenAI, the acquisition brings a new opportunity to establish a business use case for AI. That makes the acquisition interesting from a financial perspective. Many AI companies struggle with that perspective because there are not yet clear strategies to commercialize AI tools.
Attempts to be successful are often framed from use cases that are interesting to business users. They have larger budgets available to get started with AI, and companies are willing to invest because of the promised productivity improvements that AI should bring.
Also read: OpenAI considers for-profit structure to protect CEO Sam Altman
Company data shielded?
It is not yet clear when the integration will be finalized. Venkat Venkataramani, CEO of Rockset, reports that existing customers will not notice any immediate changes. Neither blog elaborates on the privacy aspect of the acquisition. Thus, it is not clear whether OpenAI’s LLMs will be trained on proffered corporate data in the future. OpenAI’s infrastructure does, however, benefit from the acquisition, according to the CEO: ”Rockset will become part of OpenAI and will drive the retrieval infrastructure that supports OpenAI’s product suite.”
For business users with paid subscriptions, OpenAI typically ensures a strict separation of corporate data. It does not enter the training data without the user’s explicit permission.