Kinetica DB Inc., the developer of a real-time analytic database for time-series and spatial workloads, today announced that it has integrated its analytic database with OpenAI LP’s ChatGPT generative artificial intelligence engine to enable users to compose SQL queries using natural language.

The company said users can now ask any question about their own data, even if those queries had not been formulated before. That’s important, it said, because conventional analytic databases require extensive data engineering, indexing and tuning to optimize query response. That means queries typically must be known in advance.

“About 99% of queries are thought out beforehand with data engineering teams,” said Chad Meley, Kinetica’s chief marketing officer. “With generative AI you’re going to see an explosion of queries that aren’t controlled.”

No pre-engineering required

Kinetica’s database doesn’t require pre-engineering of data. It uses a vectorized query engine that stores data in fixed-size blocks called vectors that can be processed in parallel. That means the query engine can operate on multiple data elements simultaneously to deliver fast results on datasets with hundreds of billions of data points, such as a recent project that created interactive three-dimensional visualizations of oil basins (pictured).

The software uses graphics processing units when available and can also run on the latest generation of central processing units, which incorporate GPUs. That has significantly reduced the cost of the hardware needed to run the database, Meley said.

“What that architecture allows us to do is lead the industry in terms of real-time analytics on both the ingest side as well as the egress side of query performance,” he said. “It also allows us to integrate more advanced styles of analytics than involve location, graph and time-series data.” Kinetica also ingests massive amounts of streaming data in real time to provide answers to questions about such things as live inventory status.

That broadens the types of questions that can be answered to include factors such as seasonality, service locations and customer relationships. The tradeoff is that Kinetica is not fully compliant with the atomicity, consistency, isolation and durability or ACID principles that ensure the data is processed reliably.

Expanding the audience

ChatGPT integration makes queries more interactive and conversational, which opens analytics to a wider audience of users and also facilitates more flexibility because data doesn’t have to be pre-optimized for a limited domain of queries. Meley called generative AI “one of the killer apps in enterprise spaces, letting people just ask natural language questions and having an analytic pop out and answer it immediately,” he said. ChatGPT’s SQL interpretation of the natural language query is displayed in advance so users can verify that the question was understood.

“The other cool thing about it is that [ChatGPT] understands your intent,” Meley said. “You can use words that don’t show up in the metadata or syntax of the schema and it will reason and figure out the correct mapping for you. The real-time nature of Kinetica will allow enterprises to not just ask what happened in the past or predict the future but also about what’s happening right now.”

Kinetica opted to license the ChatGPT from Open AI technology rather than build its own large language model. “OpenAI’s capability is pretty advanced and it would have taken us quite a bit of time to make our own base model that’s that advanced,” Meley said.

The natural language feature will be available in the free version of the company’s cloud analytics engine, which stores and analyzes up to 10 gigabytes of data. It will also be included in the paid cloud version, which costs $4.50 per hour, at no extra charge.

Kinetica, which has raised more than $77 million, said it saw 90% growth in annual recurring revenue and 163% growth in net dollar retention revenue last year. Customers include Citigroup Inc., the U.S. Postal Service, Ford Motor Co. Liberty Mutual Insurance Co. and the National Basketball Association.

Image: Kinetica