In addition to opening a world of possibilities, ChatGPT’s emergence brought to bear several underlying issues with artificial intelligence.

Those issues, which could range from data hallucination to bias, can severely affect enterprise operations. Preventing such an eventuality by streamlining data ingestion is Coda Project Inc.’s value proposition.

Coda’s Shishir Mehrotra discusses Coda Brain’s approach to tackling AI hallucination.

“Packs are the kind of extensions in Coda — they allow you to bring any other piece of information into Coda as well,” said Shishir Mehrotra (pictured), co-founder and chief executive officer of Coda. “There’s a few hundred of these that are publicly available, and then thousands more of them have been built inside companies. So, you can take Coda and you can put not only your docs, spreadsheets and so on, but all your data from Salesforce or Jira or your task-tracking systems or your analytics system.”

Mehrotra spoke with theCUBE’s Rebecca Knight and Dave Vellante at Data Cloud Summit, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed Coda’s forward-thinking approach to AI challenges, paving the way for more intelligent, efficient and integrated enterprise solutions. (* Disclosure below.)

Coda Brain: A new surface for streamlining AIOps and fostering collaboration

Coda is an all-in-one document platform that merges the functionalities of documents, spreadsheets, presentations and applications into a cohesive interface. Founded nearly a decade ago, Coda aims to eliminate the artificial divide between documents and applications, fostering a more integrated workspace.

“This idea of starting from scratch and building a single surface totally changes how teams work,” Mehrotra said. “The starting point is if you put it all in one place, you get this one home for your team to operate out of. No more copying and pasting it in different places; no more thousands of tabs to go look at. Then, people start building what we call their operating systems out of it and run planning and decision-making and your sales process and you do all your systems in it.”

AI hallucinations occur when generative systems produce incorrect or misleading information. While it might be an innocuous occurrence in a consumer context, hallucinations have extensive consequences for enterprises. Coda Brain tackles this by providing citations for unstructured data and transparent query filters for structured data, allowing users to verify and adjust the information as needed. This iterative approach ensures that the AI system evolves in accuracy and reliability, according to Mehrotra.

“We’re building it mostly with the Snowflake Cortex team,” he said. “The sandwich is, basically, we took Coda Packs, because that’s the basis of this system … that’s the brain. So, you come to Coda Brain and either the admin or the user goes and sets up whatever systems you want to integrate into the Coda Brain. One of the key things we do is we preserve all the permissions so we know who has access to every different entity. Then in the middle stack is the Snowflake Cortex stack.”

The true value of enterprise AI lies in its ability to handle complex, permission-sensitive data sets. Coda Brain’s capacity to transform queries into actionable insights and automate repetitive tasks demonstrates how AI can transcend beyond mere data retrieval to drive significant operational efficiencies, Mehrotra added.

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE Research’s coverage of Data Cloud Summit: 

(* Disclosure: Snowflake Inc. and Coda Project Inc. sponsored this segment of theCUBE. Neither Snowflake and Coda nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE