The vendor provides a marketplace for enterprises similar to those of the major cloud providers. However, it does not have its own LLM, but offers many others in its marketplace.
Dataiku on Tuesday expanded its LLM Mesh ecosystem with a governance layer.
The AI platform vendor's LLM Mesh helps IT teams build enterprise applications while also enabling organizations to adopt a multi-LLM approach by switching out various large language models to power generative AI applications, according to Dataiku.
The added governance layer helps enterprises prepare for regulatory measures and can effectively manage LLMs across organizations with an LLM registry. The registry also enables CIOs to determine which LLMs their organizations should use depending on the use case.
With the new expansion, Dataiku's LLM Mesh now supports most major LLMs, including 15 LLMs from cloud and AI vendors such as AWS, Google Cloud, Microsoft and Databricks.
The LLM Mesh expansion arrives amid new LLMs and small language models in the generative AI market. For example, OpenAI recently introduced GPT-4o mini, while Meta introduced Llama 3.1.
A separate marketplace
For Dataiku, the LLM Mesh expansion enables it to provide an LLM marketplace while helping enterprises deploy the models with features like a governance layer on top, according to Gartner analyst Arun Chandrasekaran.
This is in alignment with what we see customers are requiring.
Arun ChandrasekaranAnalyst, Gartner
"This is in alignment with what we see customers are requiring," Chandrasekaran said.
However, what differentiates Dataiku is that while it also provides models from multiple providers, it offers engineering tools like the governance layer on top of these models that enable customers to automate, customize and safely deploy the models, he added.
Dataiku's multi-LLM approach is reminiscent of a multi-cloud approach and aims to ensure organizations are not locked in to one LLM provider, said David Nicholson, an analyst at Futurum Group.
A little early
The AI platform provider was in some ways ahead of the field in providing multi-LLM access, Nicholson said.
"This Mesh idea was ahead of its time," he said. "The announcement of an enhancement is primarily a reminder to people that this is a thing."
Currently, there isn't pent-up demand for multi-LLM platforms, so Dataiku appears to be looking into the future and predicting based on what happened in the past with multi-cloud, and trying to get ahead of the issue, Nicholson continued.
"They're going to almost be waiting for people to get to the point where they've figured out what to do with generative AI in a business context, trying to get them early to say, 'Listen, before you try, before you do anything, put a layer in so you don't become captive to an individual company's set of tools,'" he said.
The challenge for Dataiku is that as it adds new features like governance into the LLM Mesh, it will need to keep up with other updates in the LLM market as well, Nicholson added.
"Sometimes when you put an intervening layer in, you're increasing complexity, but you could be decreasing performance," he said.
Esther Ajao is a TechTarget Editorial news writer and podcast host covering artificial intelligence software and systems.
