ServiceNow’s AI Experience is an agentic AI UI for the Now Platform

ServiceNow today launched the AI Experience (AIx), a contextually aware multimodal AI-driven user interface (UI) for its Now platform. Building on the ServiceNow AI Platform and with a foundation in Now Assist, the company describes it as “a unified, conversational front door to enterprise AI.”

With AIx, intelligent role-aware AI agents work alongside humans to perform tasks such as evaluating trends, triaging support issues, suggesting problem resolutions, or generating price quotes, across the entire ServiceNow platform, the company said.

It all is managed, monitored, secured, and governed via the recently announced AI Control Tower.

New capabilities include AI Voice Agents, which enable users to retrieve information, update records, and troubleshoot issues using voice commands; AI Web Agents, which learn from humans how to navigate third party apps and the web, and how to fill out online forms without the need for integrations or APIs; AI Data Explorer, which uses Workflow Data Fabric to connect data from ServiceNow and external sources to help users investigate trends and perform other tasks without leaving their workflows; and AI Lens to interpret visual input, such as a screenshot of an error message, and provide answers to questions about it in plain language.

The company expects to make the voice and web agents, AI Data Explorer, and a new AI-powered Configure, Price, Quote (CPQ) sales tool available by the end of 2025. AI Lens is generally available today.

One platform

The new AIx supports all users of the platform, including employees, knowledge workers and developers, said Amy Lokey, EVP and chief experience officer at ServiceNow, during a media briefing. “Within the platform, we have AI native workflows and AI agents. We also have real time integration into other systems of record to drive outcomes, and we have an enterprise grade foundation with trusted security, scalability and governance built in.”

Lokey noted that, for example, the AIx will streamline discovery of issues by enabling support managers to simply ask an agent to analyze critical incidents across text, voice, and web inputs to spot trends, flag inefficiencies, and suggest fixes or process improvements.

“It’s one platform, one system of action, one unified intelligent experience for enterprise AI,” she said.

In addition, said Dorit Zilbershot, group VP product management, AI Experience and Innovation, it enables customers to combine general purpose models with domain-specific intelligence.

“The future is about moving from agents that just access data to agents that actively collaborate, delegate, and perform tasks across the enterprise,” she said. The goal is to provide “a holistic operating model for agent interoperability, helping customers connect, orchestrate, and govern their AI agents, regardless of vendor and department. We want to make sure that whether it’s MCP, A2A, or any other agent-to-agent protocol, we hide the complexity, giving our customers the optionality to choose how their AI agents will collaborate across the enterprise.”

Every AI-driven action is logged, explained, and auditable, she said. To that end, AI Control Tower is adding agent autodiscovery to its bag of tricks, showing which models are in use as well as their risk and compliance status.

On the CRM front, ServiceNow is using AIx to transform its product into what it calls Autonomous CRM, in which AI agents take on repetitive tasks such as scanning tickets, flagging patterns, and recommending response plans, enabling the human agents to focus on complex decisions. The new CPQ component, for example, speeds the generation of customized price quotations.

Clearing the cache of bad training data

Brian Eble, principal, enterprise technology in the core business services technology group at EY, has recently launched a ServiceNow virtual assistant for IT support and said that it has significantly reduced the number of tickets directed to human agents.

His biggest challenge was the quality of the data used to train the AI agents. Knowledge bases had to be cleaned up, and obsolete information removed. For example, he noted, many web app troubleshooting responses include the requirement to clear the browser cache. “We’ve written how to clear your cache or when to do it and how to do it, 50, 60 different ways,” he said. “Therefore, the AI is challenged to always understand that this is actually the same thing, and that we want it to give a consistent response.”

The value of AIx not just about passing tasks to an AI, he added. “People kind of forget that sometimes helping our live agents is a very powerful, impactful thing, too. So if they can answer questions faster, more accurately, then that’s a good thing. Wait times are lower or non existent. You can do more with the same amount of people.”

AI is the new UI

Analysts see the AIx as yet another indication of a shift in the market.

Thomas Randall, research director at Info-Tech Research Group, said, “ServiceNow is positioning itself to integrate customer-facing and operational workflows with a single, AI-based layer of engagement.”

Phil Carter, group VP/GM, AI, Data and Automation research at IDC, agreed.

“AI is becoming the new UI,” he said. “Increasingly, users will interact with enterprise data and the associated business logic through non-deterministic, often conversational, means. This is a really important move from ServiceNow to capture the ‘enterprise eyeball’ — it will be critical for stickiness with knowledge workers, and the usage data from AIx will drive the next wave of AI innovations.”

The longer-term value, he said, will be in combining ServiceNow’s AIx and AI Control Tower: “You need the innovation at the UI level to be combined with governance, workflow orchestration as well as integration/interoperability with third party environments, increasingly via A2A and MCP, via the AI Control Tower.”

However, with competition in the space ramping up, UI/UX and agentic sprawl could be a concern, he said, noting that Workday’s acquisition of Sana adds an AI UI, Microsoft has been pushing Copilot, and IDC expects that Slack will become the primary UI for Salesforce applications over time. CIOs, he said, need to be “very deliberate” about which vendors they chose to partner with for agentic AI.