Large language models such as Google LLC’s Bard and OpenAI LP’s ChatGPT are underpinned by generative artificial intelligence, and several companies have understood gen AI’s potential to transform business.

One of those companies is ChaosSearch Inc., which has adapted the technology to its platform and offers an AI assistant capable of understanding data.

“As a founder, as a technical founder of an analytical company, I said, ‘We’ve got to be part of this,'” said Thomas Hazel (pictured), founder and chief technology officer of ChaosSearch. “Early on, we adapted the technology to ChaosSearch, and we offer an AI assistant that you can now have a conversation with your data. It demos really well, but it really solves the problem of ‘I don’t understand complex APIs. I don’t understand high-level tooling.’ I can just log in and have a conversation just like you do with ChatGPT.”

Hazel spoke with theCUBE industry analyst Dave Vellante at Supercloud 4, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed AI, data quality and the efficient deployment of AI in data analytics.

ChaosSearch’s role in the data revolution

As a data analytics company, innovation for ChaosSearch means finding new ways to streamline the extraction of insights from all kinds of enterprise data — AI is the next frontier in that regard. ChaosSearch focuses on log analytics at scale, facilitating data integration and analysis. The company recently unveiled Chaos LakeDB, enabling customers to stream data into their data lake while auto-discovering and indexing it.

“We took our data solution, and we integrated well-known LLMs, like ChatGPT and Vertex AI, and now we’re working with Amazon’s LLMs,” Hazel said. “And what we’ve done is use it as a reasoning engine. So, we don’t put data into the LLM; we actually ask the LLM intelligent questions to ask on our database.”

This approach allows for compatibility with familiar application programming interfaces such as Elastic API and SQL via Trino, eliminating the need for costly and complex tools, Hazel added.

“For instance, I might say I’m a security analyst and I present the LLM with, say, a basic schema and ask it to craft a search query and or a SQL query to do the query on our database,” he explained. “It reduces the hallucination. It obviously is dramatically less expensive to build, to fine-tune your own LLM. That combination allows our customers to interact in a new way, a conversational way.”

Democratizing data access

One of the most transformative aspects of ChaosSearch’s approach is its commitment to data democratization. By allowing users to interact with their data in a conversational manner, they eliminate the need for in-depth knowledge of complex data tools or APIs. This is a game-changer for those who have valuable data but lack the technical expertise to fully harness it, according to Hazel.

“What’s interesting is that prompt engineering really goes across all the LLMs that we’ve been using,” he said. “Now, different LLMs are better than others. ChatGPT is pretty good. Vertex AI is pretty good. Bard is pretty good. But as they get better, our reasoning engine gets better. So, to us, we have a plug-and-play architecture that brings in the LLM with this chain of thought, with this prompt engineering we do and, again, our backend high-scale database.”

This approach extends beyond businesses and into educational settings, raising the question of whether prompt engineering, the process of crafting effective questions for AI, should be taught in schools. It is clear that AI and data are becoming increasingly relevant in our daily lives, and understanding how to leverage these technologies is a crucial skill.

Log analytics is quietly booming right now, as evidenced by Cisco Systems Inc.’s acquisition of Splunk Inc. for around $28 billion. For its part, ChaosSearch aims to capitalize on these market trends by offering a more modern and cost-effective alternative to traditional log analytics platforms, according to Hazel.

“What we want to do is go after those use cases but with modern architecture, a lake architecture — and doing it in a way that dramatically reduces the cost,” he said. “Very often we replace Splunk or augment Splunk because great product, but at scale, can cost you millions, tens of millions of dollars.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Supercloud 4:

Photo: SiliconANGLE