ServiceNow AI Control Tower wants to govern agents on every platform

ServiceNow is not an iPaaS vendor, but its announcement at Knowledge 2026 on 5 May touches the same question integration teams are working on: who keeps track of all the AI agents in a company when they run on many different platforms? ServiceNow’s answer is AI Control Tower, which it now extends to AI that runs outside ServiceNow.

What ServiceNow launched

ServiceNow organises AI Control Tower around five tasks: discover, observe, govern, secure and measure. The main additions are:

• Discovery across platforms. Thirty new integrations find AI assets in AWS, Google Cloud and Microsoft Azure, and in applications such as SAP, Oracle and Workday. Discovery also covers non-human identities and connected devices.
• An AI gateway for MCP. A new gateway gives real-time control of Model Context Protocol traffic, with governance, observability and security for third-party AI systems.
• Runtime observability. ServiceNow has bought Traceloop to see what agents actually do while they run.
• Identity. An integration with Veza covers access governance for AI in the cloud platforms.

The enhancements roll out gradually. According to ServiceNow, the new AI Control Tower features go into its Innovation Lab in May, with general availability expected in August 2026.

Why it matters

The pattern is becoming clear. Every large platform now wants to be the place where a company registers, watches and controls its agents. Cloud providers, iPaaS vendors, API management vendors and now ServiceNow all offer some form of agent registry, AI gateway and identity for agents.

ServiceNow’s angle is that it already owns much of the IT operations and IT asset data in large companies. Treating agents as assets, alongside servers, applications and licences, fits that position. The discovery integrations are the key part: a control tower that only sees one platform does not answer the question it was built for.

How it compares

For integration teams, the overlap is in the AI gateway and the registry. An iPaaS with its own MCP gateway, an API gateway with MCP support and ServiceNow’s AI gateway may all sit in the same company. They can complement each other: the integration platform controls traffic to business systems, while ServiceNow gives management a cross-platform overview. They can also compete for the same role.

The risk is that each tool sees only the traffic that passes through it, and that no one has the full picture.

What to ask

1. What is the source of truth? Decide which tool is the main inventory of agents and MCP servers, and which ones only report to it.
2. Where does MCP traffic go? Map which gateway each type of MCP traffic passes through, and make sure there are no paths without logging.
3. Discovery or registration? Ask whether the tool finds agents by scanning other platforms, or only knows about agents someone has registered.