ServiceNow bets the enterprise on an autonomous workforce
At Knowledge 2026 the company put AI specialists on every business function and a control tower over all of them. The pitch is aimed at the pilot-to-production wall.
- At Knowledge 2026, ServiceNow expanded its "Autonomous Workforce" of AI specialists that complete whole processes rather than just assist, across every major business function.
- It broadened its AI Control Tower to discover, govern, secure, and measure AI running anywhere in the enterprise, with a new layer aimed at tracking cost and return.
- Named results: the City of Raleigh cut IT service desk costs by 66% with a 98% deflection rate, and Honeywell reported 75% faster compliance attestation.
- A separate ServiceNow and Accenture program puts engineers inside customer environments to move agents from pilot to production.
ServiceNow used its Knowledge 2026 conference to make the case that the enterprise is ready to hand real work to agents, not just test them. The centerpiece was a broad expansion of what the company calls the Autonomous Workforce, a set of AI specialists meant to run entire business processes end to end. ServiceNow brought that workforce to every major function, from IT and HR to finance and customer operations.
What ServiceNow announced
Alongside the agents came the thing meant to keep them in line. ServiceNow expanded its AI Control Tower to discover, observe, govern, secure, and measure AI across any system in the enterprise, not only its own. The measure part matters most for buyers. It adds cost tracking and return dashboards aimed at what the company describes as one of the biggest problems in enterprise AI, runaway model spend. As Fortune reported from the event, the framing was blunt: enterprises need AI that senses, decides, and acts, but they also need to see what it costs.
The numbers behind the pitch
ServiceNow leaned on customer results to sell the shift. The City of Raleigh runs a virtual agent that deflects 98% of employee requests and cut IT service desk costs by 66%, according to the company's published customer story. Honeywell reported 75% faster compliance attestation and said the agent eliminated most of its service desk conversations. Numbers like these are early and vendor-sourced, so weigh them accordingly, but they point at where agents are already earning their keep, in high-volume, repetitive workflows.
The real target: pilot to production
The whole push aims at the wall most enterprise agent projects hit, the gap between a promising pilot and a production rollout. ServiceNow and Accenture also launched a Forward Deployed Engineering program that embeds engineers inside customer environments to build and scale agents where the work already runs. The bet is that governance, measurement, and hands-on help are what turn a demo into a deployment. For how these agents coordinate once they are live, see our section on orchestration, and for the wider platform field, our ranking of agentic AI platforms.
Sources: ServiceNow newsroom, Fortune, ServiceNow City of Raleigh story.