ServiceNow’s annual Knowledge conference opened this morning at the Venetian Resort in Las Vegas with a theme that could not be more direct: “Put AI to Work: The Agentic Era.” Running through May 7, the event brought together tens of thousands of enterprise practitioners to watch what may be the clearest signal yet that AI is moving from co-pilot to employee.

NVIDIA CEO Jensen Huang joins ServiceNow chairman and CEO Bill McDermott on the main stage today. The headline product: an AI specialist that does Level 1 IT support, end to end, without a human in the loop.

The Australia Release Goes Live Today

ServiceNow’s Australia platform release reached general availability today, coinciding with the conference. The release marks what ServiceNow is calling a structural transformation of the platform’s core architecture, moving from feature additions on top of an existing foundation to something they describe as an AI-first agentic architecture.

The practical difference is significant. Previous releases added AI features alongside existing workflows. Australia rebuilds the underlying execution model so AI agents can plan, decide, and complete work independently, not just assist a human who is doing the work.

Several specific capabilities ship with Australia:

AI Control Tower is the governance layer. It gives IT and compliance teams a centralized dashboard to monitor every AI agent running in the enterprise, track adoption, manage costs, and enforce guardrails. Notably, ServiceNow describes the governance model as shifting from “observe and report” to “control and enforce.” In practical terms: unapproved AI configurations are blocked, not just flagged. For organisations that have been worried about shadow AI proliferating without oversight, this is the enforcement mechanism that was missing.

AI Agent Orchestrator allows specialised agents to coordinate across departments to solve complex issues together. A workflow involving HR, IT, finance, and legal no longer needs a human to hand off between teams. Agents can route, escalate, and hand off to each other.

AI Agent Advisor identifies and prioritises where deploying an AI specialist would generate the most measurable business value before committing to deployment. It is meant to prevent the proliferation problem where AI agents get built because someone thought it would be useful, with no baseline measurement.

AI Desktop Actions watches how users actually complete tasks and uses those patterns to train agents. Instead of documenting processes for AI to follow, the system infers them from real behaviour.

RaptorDB Pro underpins all of it. ServiceNow’s high-performance database engine is specifically designed to handle the data throughput required when AI agents are executing decisions at transaction volume, not just surfacing recommendations.

The platform also becomes model-agnostic in Australia. Organisations already running OpenAI, Anthropic, or their own fine-tuned models can connect them into the ServiceNow orchestration layer without rearchitecting.

The First AI Specialist: Level 1 Service Desk

The announcement that will resonate most with operations teams is the Level 1 Service Desk AI Specialist, reaching general availability in Q2 2026. It handles the work that floods most corporate IT queues: password resets, software access provisioning, and network troubleshooting.

The agent does not just log a ticket. It accesses enterprise knowledge bases, checks historical incident data, runs the fix, and confirms resolution. It escalates to a human only when something falls outside its defined scope. ServiceNow says it operates 24 hours a day, seven days a week, with work assignments aligned to specific skillsets and governance rules.

That last point matters for enterprise adoption. Autonomous execution sounds efficient until a compliance officer asks what authority the agent had to provision access to a production system at 2am. ServiceNow’s answer is that every action is governed by the same role-based permissions a human employee would have, with a complete audit trail.

The Moveworks acquisition, which closed in December 2025 for $2.85 billion, shows up here as ServiceNow EmployeeWorks. The product combines Moveworks’ conversational AI and enterprise search with ServiceNow’s portal and automated workflows. Employees make requests in plain language; the system converts them into governed, end-to-end actions. ServiceNow says EmployeeWorks is designed to serve approximately 200 million employees globally.

The NVIDIA Layer

Jensen Huang’s presence at the Knowledge 2026 keynote reflects a partnership that has been building since NVIDIA’s GTC conference in March. The integration brings NVIDIA’s Agent Toolkit, Apriel models, and Nemotron models into ServiceNow’s orchestration layer, running on NVIDIA’s Blackwell AI infrastructure.

The practical outcome is that ServiceNow customers can now deploy agents that leverage NVIDIA’s inference capabilities and models, including open and closed options, without building a separate AI infrastructure stack alongside the ServiceNow platform. The AI Control Tower and NVIDIA’s AI Factory are being designed to work together on governance, so organisations managing agents built on both platforms can see everything from one place.

What This Means for Business

Two major enterprise AI platforms, ServiceNow and IBM (which held Think 2026 in Boston this week), are saying the same thing simultaneously: the pilot era is ending.

For business leaders still running AI proofs of concept, the message is uncomfortable. The tooling has matured faster than most organisations have moved. A Level 1 IT specialist that works around the clock and logs everything is no longer a research project. It is a product you can deploy in Q2 2026.

The implications are not primarily about headcount reduction, though that conversation is unavoidable in any back-office context. The more immediate implication is operational capacity. Most IT teams are bottlenecked not by skill but by volume. The number of access requests, password resets, and routine troubleshooting tickets grows faster than teams do. An agent that absorbs that volume changes what the human team does with its time.

The governance architecture in the Australia release is also worth noting separately from the automation capabilities. The AI Control Tower’s shift from flagging to blocking unapproved configurations is a meaningful change for organisations in regulated industries. Healthcare, financial services, and government entities have been cautious about agentic AI precisely because the accountability trail was unclear. ServiceNow is making the audit trail structural, not optional.

For teams evaluating where to start with agentic AI, the IT service desk is a high-return, low-risk entry point. The scope is defined, the success metrics are clear, and the compliance requirements are well understood. It is also, notably, where ServiceNow is putting its first out-of-the-box specialist.

Whether you are running ServiceNow today or not, the broader pattern is worth watching. The enterprise software vendors that matter most are competing to become the orchestration layer where AI agents operate. The decision about which platform handles that role inside your organisation is one of the more consequential infrastructure decisions of the next two years.