Kinetica leverages ChatGPT for natural language SQL database queries

Kinetica, the relational database provider for Online analytical processing (OLAP) and real-time analytics OpenAI ChatGPT Make it available to developers natural language processing Execute a SQL query.

Kinetica, which offers databases in multiple flavors including hosted, SaaS, and on-premises, announced Tuesday that it will offer ChatGPT integration for free with its free developer edition, which can be installed on any laptop or PC. Added.

The ChatGPT interface built into Kinetica Workbench’s front end can answer queries asked in natural language about proprietary data sets in the database, the company says.

“ChatGPT brings natural language to Structured Query Language (SQL)Therefore, users can enter and submit arbitrary queries. APIs Quit ChatGPT. In return, you get SQL syntax that you can run to generate results,” said Philip Darringer, vice president of product management at Kinetica.

“Furthermore, we can understand the intent of the query. This means that the user does not need to know the exact names of the columns to query. We map to columns, which is a big step forward,” added Darringer.

To make inferences clear from natural language queries, Kinetica’s product managers created some prompts and context based on their knowledge of the database already deployed in ChatGPT.

“We send specific table definitions and metadata about our data to our generative AI engine,” Darringer said, adding that corporate data is not shared with ChatGPT.

The database continuously ingests streaming data, according to the company, so it can answer even the latest real-time analytical queries.

Vectorization speeds up query processing

According to Kinetica, vectorization improves the speed at which relational databases process queries.

“In a vectorized query engine, data is stored in fixed-size blocks called vectors, and query operations are performed on these vectors in parallel rather than on individual data elements,” the company said. I added that this allows the query engine to process it. Achieve faster query execution with a smaller compute footprint by processing multiple data elements simultaneously.

Kinetica uses a combination of graphical processing units (GPUs) and CPUs to enable vectorization, the company said, adding that the database uses SQL-92 as its query language. PostgreSQL and MySQLand text search, time series analysis, location intelligence, and graph analysis, all now accessible in natural language.

Kinetica claims that ChatGPT integration makes the database easier to use, increases productivity, and improves insights from data.

Bradley Shimmin, Chief Analyst at Omdia Research, said:

According to Shimmin, Kinetica was one of the first database companies to integrate ChatGPT or generative AI capabilities within its database.

“But within the databases themselves, there hasn’t been much effort to integrate natural language query (NLQ). It is used by other practitioners who are familiar with the work,” said Simin. Business Intelligence (BI) The market has made more progress in NLQ integration.

According to Shimmin, Kinetica’s use of ChatGPT for natural language queries is “clever”, but strictly speaking, it’s not a real database query.

“Kinetica is not talking about using natural language to query databases. Rather, Kinetica works like Pinecone, Chroma, and other vector databases, providing searchable We create an index (a vectorized view) and feed it into a natural language model such as ChatGPT to create a natural way to search the vectorized data.

“One very popular implementation of this kind of conversational query is a combination of Chroma, LangChain and ChatGPT,” Shimmin added. Chroma is an open source database and LangChain is a software development framework.

However, Shim believes the integration will be “very” favorable to Kinetica.

“As enterprise practitioners start looking for ways to deploy large-scale language models (LLMs) that work behind firewalls without spending a lot of money training their own LLMs or fine-tuning their existing databases. , the vector database will be a hot ticket in the second half of 2023. LLM uses company data,” said Shimmin.

Kentica said it is open to working with other LLM providers as new use cases arise.

Chad Meley, Kinetica Chief Marketing Officer, said:

The company, which derives more than half of its revenue from US defense agencies such as NORAD, has customers in the logistics, financial services, telecommunications and entertainment sectors, as well as in the connected car space.