Getty Images/iStockphoto

DataRobot GenAI features aim to address enterprise concerns

The AI vendor's new capabilities addresses concerns about the new technology, such as governance, cost and scale. It also helps enterprises bridge two types of AI.

DataRobot on Thursday introduced new governance and observability capabilities for generative and predictive AI technology for its AI platform.

The capabilities include Generative AI Guard Models, LLM Cost and Performance Monitoring, Multi-Provider LLM Playground, Unified AI Registry and Generative AI Accelerators.

The new features come after the AI vendor updated its AI Platform 9.0 in August with generative AI-specific capabilities that enable enterprises to build, deploy, manage and monitor their generative AI assets such as large language models (LLMs).

The governance issue

It also comes as many enterprises are looking to deploy generative AI (GenAI) models safely after identifying appropriate applications and model providers, said Gartner analyst Arun Chandrasekaran.

As enterprises start leveraging large language models within their environment, they recognize that they need to have governance around these models.
Arun ChandrasekaranAnalyst, Gartner

"As enterprises start leveraging large language models within their environment, they recognize that they need to have governance around these models," Chandrasekaran said.

One of the primary governance goals is ensuring that the models are not hallucinating or mistakenly revealing personal information.

"One of the biggest fears that enterprises have of using generative models is that it will [reveal] something that they're not comfortable with or that their customers won't be comfortable with," said Forrester Research analyst Mike Gualtieri.

One of the new features, Generative AI Guard Models, is designed to help with this challenge.

The feature lets enterprises create another AI model that looks at the LLM output and assesses it for errors such as toxicity and leaks of personally identifiable information.

"What these tools essentially do is to provide that visibility and provide control to the extent possible within the environment," Chandrasekaran said.

But these governance tools, while helpful, are not likely to fix all the problems enterprises have with generative AI technology, he added.

Predictive AI and generative AI

Other than governance, DataRobot is aiming to help organizations bridge the gap between predictive AI and generative AI.

Unified AI Registry is a system of record for enterprises to govern all their generative and predictive AI systems.

This is especially important for enterprises that became introduced to AI technologies through generative AI, said Mike Leone, an analyst at TechTarget's Enterprise Strategy Group.

"It's really being able to bridge the gap for those new organizations to understand, 'hey, there's a lot more in this AI realm than generative AI,'" he said.

Generative AI and predictive AI also complement each other. Meanwhile, generative AI will be helpful to enterprises primarily focused on predictive AI, Gualtieri said.

For example, a telecom company that previously used predictive AI to forecast customer churn can now use generative AI to write personalized emails to customers whom its predictive models said are likely to churn.

Cost transparency

Another enterprise worry DataRobot is looking to tackle with its new capabilities is cost.

The LLM Cost and Performance Monitoring capability observes and manages an organization's generative AI systems and costs. It allows organizations to see the cost of their generative AI projects and set alerts to avoid going over budget.

Due to the newness of GenAI, cost transparency is needed within the market, according to Leone.

"Organizations still don't have a lot of understanding on cost," he said.

There are times when organizations are excited about a generative AI application and become discouraged because the amount of money needed to start the AI project and the ongoing costs don't fit their budget.

"As organizations are looking to really understand costs so they can project what value will look like, they need a level of insight," Leone said, adding that DataRobot is taking steps to provide this ability to estimate AI cost.

Tool comparison and scale

DataRobot also is aiming to address enterprises' concerns about using the best tools for their application.

Multi-Provider LLM Playground provides users access to Google Cloud Platform, Vertex AI, Azure OpenAI and AWS Bedrock so enterprises can compare and experiment with any combination of foundation models, vector databases and prompting strategies, according to DataRobot.

This tool will help enterprises deal with the innovations in generative AI models and test them, Gualtieri said.

Enterprises are also looking for ways to scale their AI projects, he added.

DataRobot's Generative AI Accelerators enables enterprises to quickly deploy their generative AI projects and build retrieval augmented generation applications, the vendor said.

With the new features, DataRobot is seeking to enable enterprises to move from one end of the generative AI lifecycle to the other -- from ideation to deployment. But the AI platform provider is among several independent vendors and tech giants trying to do the same thing.

Last month, Domino Data Lab, a DataRobot competitor, updated its AI platform with new features to accelerate generative AI projects.

Now, as vendors like DataRobot, Domino Data Lab and Dataiku provide AI platforms that have looked to help enterprises move from one end of the AI lifecycle to the next -- experimentation to deployment -- their pivot into generative AI ought to have enterprises drawn to the features they provide, Gualtieri said.

"The AI platform market will take on GenAI as yet another form of AI," he said. "These vendors have added features that are GenAI-specific. They've got a well-thought-out strategy for how they're going to accommodate it."

Esther Ajao is a TechTarget Editorial news writer covering artificial intelligence software and systems.

Dig Deeper on AI technologies

Business Analytics
Data Management