ServiceNow is fully embracing agentic AI with a series of new product launches announced during Knowledge 2025 in Las Vegas.
The ServiceNow AI Platform has been built for “the agentic and open era of AI”.
ServiceNow chairman and CEO Bill McDermott said the product utilises AI in a way where it “isn’t just another tool, it’s deeply embedded in the way work gets done”.
The new platform unifies intelligence, data and orchestration, enabling enterprises to transition from “fragmented pilots to full-scale AI execution”.
This is achieved via an AI Engagement Layer, which ServiceNow said will allow customers to complete complex tasks across systems by using its Knowledge Graph, Workflow Data Fabric and AI Agent Fabric solutions, the latter of which was also launched at the conference to facilitate agent‑to‑agent and multi‑model communication and collaboration.
ServiceNow president, chief product officer and chief operating officer, Amit Zavery said the platform was built to be open from day one.
“For AI to be truly transformative, it can’t work in isolation. It needs to be deeply embedded in the way work gets done. You choose your model, your governance, your region, from our own domain specific models to popular models,” he said.
This includes GPT-4, Gemini, NeMo, Titan, Claude, and other open source models like Llama and Mistral.
This isn’t just relegated to models but infrastructure as well, as it is compatible with ServiceNow’s cloud as well as Azure, GCP, private cloud, government sovereign clouds and even on-premises solutions. Additionally, users can burn down their hyperscaler cloud commits using ServiceNow through hyperscaler marketplaces.
Existing ServiceNow customers can utilise the platform from today, with no migrations or disruptions, Zavery added.
“We have already built thousands of AI agents, ready to deliver immediate value on day one. They were developed to support every industry across every corner of your business, and this includes IT, CRM, HR, finance, legal, application development and security,” he said.
McDermott also shared his view on the future of AI, explaining how he believes that traditional app stacks are set to collapse.
He predicted the number of apps as they are known today will be “radically reduced” in the future.
“21st century problems cannot be solved with 20th century architectures,” he said.
Instead, he suggested that systems from the 20th century will become core databases that will feed into ServiceNow’s platform, where “the real work is happening”.
“AI can only achieve its full potential when it spans across every system, every department and every function,” he said.
ServiceNow also took the wrappers off its AI Control Tower, which can manage, govern and secure AI agents – not just those from ServiceNow but any vendor — across an enterprise from a single command centre.
The ServiceNow AI Control Tower is generally available, while AI Agent Fabric is available for early adopters and is planned to enter general availability in Q3.
ServiceNow CRM was also launched which is integrated with AI and aims to work across departments for “for fast, seamless experiences across the entire customer lifecycle, reducing time spent swivelling between applications and customer requests”.
Alongside this was the launch of AI agents for CRM, which can complete tasks autonomously, scale call centres and sales teams, as well as making “true self-service a reality”, ServiceNow added.
Both ServiceNow CRM and CRM AI agents are generally available now and can be accessed via the ServiceNow Store.
Additionally, cross compatibility of AI agents between Copilot Studio and Azure AI Foundry, which allows the two to “dynamically exchange information, coordinate tasks and take action in real time”.
“This new capability will accelerate an agentic ecosystem that helps unlock new levels of employee productivity,” Microsoft CEO Satya Nadella said via a prerecorded video. “Of course, this is just the beginning.”
ServiceNow is also stepping up to educate users via ServiceNow University, a free gamified experience containing hundreds of on-demand courses and hands-on performance-based assessments that provide credentials and points, badges and rewards with an AI-powered profile to curate content.
Apriel Nemotron 15B was also released, which is a reasoning model developed in collaboration with NVIDIA that can evaluate relationships, apply rules and weigh goals to reach conclusions and make decisions. It will be available during Q2.
The open-source large language model (LLM) has been post-trained with NVIDIA and ServiceNow-provided data to help deliver lower latency, lower inference costs and faster agentic AI.
Sasha Karen travelled to Knowledge 2025 as a guest of ServiceNow.
