ServiceNow launches enterprise AI governance capabilities

)

ServiceNow’s flagship Now platform has been a cornerstone for enterprise IT operations for many years.

Over the past few yearsthe company has grownits capabilities by introducing enterprise AI capabilities. This includes Now Assist. It is critical that an organization’s platform has a high level of confidence, as it is used to run their operations. Enterprises have expressed concerns about the safety of generative AI and possible hallucinations.

The company announced today a series new governance capabilities for the flagship Now platform, designed to increase confidence in enterprise AI use. The new governance features address an increasing challenge in enterprise AI adoption, namely the gap between full production deployment and experimentation.

Governance components include Now Assist Guardian and Now Assist Data Kit. The new tools allow organizations to manage AI deployments throughout their enterprise. These tools are essential as companies move from proof-of concept stages to full production environments.

Jeremy Barnes, VP AI Product, ServiceNow, told VentureBeat that last year was more of an experimentation phase, and this year is more real. “People are deploying AI to improve their bottom line or their top line.”

Why AI governance is important for enterprise adoption

Governance and compliance are key operations in an enterprise.

ServiceNow recognizes the complex relationship between enterprise stakeholders. Barnes said that “typically, our customers have governance and compliance within a different organization than the organization that is defining and owning economic benefits of generative AI.”

In most organizations, this means that one team is able to put together a proof-of-concept in order to test generative AI. The constraints are different at this stage than when an application or service has been rolled out to an entire enterprise as part of a full-scale production deployment. A governance team will inevitably tell the development team they cannot deploy anything without first ensuring that it is compliant with the organization’s policy. Barnes explained that this leads to generative AI efforts being stuck in ‘limbo,’ between proof-of-concept and production for an extremely long time.

Barnes noted that the AI governance updates are helping to bridge this divide, by providing tools and transparency that satisfy business and compliance needs. Barnes said that “AI governance is more than just researching models.”

Barnes explained that it is about having a system with AI components and traditional work flows. It’s all about understanding the system and being able ensure that it fits the desired outcome of the enterprise. Governance is also about knowing when something goes wrong and being able to manage the situation.

How agentic AI accelerates governance imperative

Agentic AI is being deployed, and this is one of the reasons that more AI governance is required now.

Many companies, including ServiceNow are deploying agent frameworks in order to give AI more autonomy. Barnes stated that with more autonomous AI systems, there is an increased need for robust governance and controls, as well as human oversight, to ensure that the systems are operating within acceptable parameters and according to their intended purpose.

ServiceNow’s governance tools and workflows are designed to help enterprises manage risks and maintain the required level of control over more autonomous AI systems.

The intersection of enterprise AI Governance and Hallucination

One of the biggest challenges for enterprise adoption is hallucination. Governance is not the solution to this challenge, but is a necessary component.

The issue of hallucination affects all generative AI models. ServiceNow has a multi-layered strategy to mitigate hallucination. The approach includes fine tuning language models so that they are more focused on extracting data than generating new data.

Governance plays a critical role in mitigating risk. Now Assist Analysis Guardian will provide an additional layer of protection by analyzing AI outputs. Barnes said that ServiceNow’s goal is to ensure that hallucination does not become a’showstopper,’ but is instead viewed as a potential risk that can easily be addressed by the platform’s tools.

How enterprise AI can help Configuration Management Database deployments.

Configuration Management Database is a cornerstone in IT operations management. CMDB systems keep track of the software, hardware and configurations that are used in an organization.

As a part of today’s ServiceNow update, there is a new Now Assist CMDB feature that brings the power AI. Barnes explained that this new capability is not aimed at directly populating or discovering the CMDB as it is usually done by other means, but instead focuses on improving productivity for users interacting with CMDB data.

ServiceNow’s broader AI-powered productivity strategy includes the CMDB Analysis feature. The CMDB Analysis feature is integrated into the AI Governance Framework, ensuring that deployment and use of the AI-powered tool are subject to the same governance and controls. This helps IT operations teams overcome the trust and operational concerns they may have when deploying AI tools in their critical systems and data.

Barnes said, “The more you rely on AI tools, the more you have to make sure they are trustworthy for your work.”