Eleven companies joined ServiceNow on Wednesday as founding members of the Control Tower Alliance, a partner program built around the Super Intelligence Control Tower, the product ServiceNow sells for finding, governing, and measuring the artificial intelligence systems a company runs. The members are Anthropic, Amazon Web Services, Cisco, HiddenLayer, Kong, NVIDIA, Okta, OpenAI, Stripe, WitnessAI, and Zenity.
Partners will build solutions that run on the Control Tower and pass ServiceNow qualification, certification, and integration testing before customers see them listed. The launch release also commits partners to shared standards for agent identity and safety guardrails, and it says ServiceNow will sell and support selected joint offerings worldwide. The program runs alongside the ServiceNow Store and the Workflow Data Network Partner Passport.
In July I wrote a primer on why ServiceNow, Microsoft, and Boomi are each building a control tower, and in August I covered how ServiceNow rolled six security products into one platform. This announcement puts named partners on the work that happens while an agent is running.
ServiceNow Keeps the Policy Record While Partners Handle Live Traffic
The Control Tower holds the organizational view: which agents and models exist, who owns each one, how risky each is, and what the business has approved them to do. Amit Zavery, ServiceNow's president, chief product officer, and chief operating officer, said in the launch release that no single solution "can see everything," and that AI governance needs an ecosystem of partners. Each partner adds visibility or control from a position ServiceNow does not occupy, such as the network path, the identity provider, or the model itself.
For the network and security teams who will run this, Cisco and Kong matter most, because their products sit closest to the assets and the calls agents make.
Cisco Brings Integrations You Can Install Today
Cisco's contribution is AI Defense. In a post on Cisco's blog, product leader Spencer Colemere described four functions:
- Discovering models, agents, datasets, and Model Context Protocol servers across cloud environments
- Scanning models and tool servers for risks such as malicious code before teams adopt them
- Red-teaming models and applications for prompt injection and data leakage, before launch and in production
- Applying security, privacy, and safety guardrails to applications and agents while they run
Cisco also has a head start on availability. Two integrations from the partnership Cisco and ServiceNow expanded in April 2025 are already on the ServiceNow Store (Cisco, 2026). The first feeds assets that AI Defense discovers into the Control Tower inventory. The second sends AI security detections into ServiceNow AI Security Exposure Management, where your team assigns an owner and tracks the fix.
Cisco remains the source of the security finding, and ServiceNow holds the ticket, the owner, and the business service that depends on the affected model.
Kong Offers One Runtime Policy Across Model Providers
Kong joins with Kong AI Gateway, part of its Konnect platform, which handles three kinds of traffic: calls to large language models, calls to tools over Model Context Protocol, and handoffs between agents. In Kong's launch post, head of product marketing Alex Drag described what the gateway does on each request. It identifies the calling agent and the person it acts for, checks the request against policy, screens prompts and responses for sensitive data, caps spending before a budget runs out, and meters tokens back to the team and application that used them.
Kong's argument rests on neutrality, because enterprises run agents on several clouds and model providers at once and each provider governs its own traffic. Kong calls its answer strategic portability, meaning one runtime policy that stays in place when you switch model providers. If your team expects to change model vendors during the life of this contract, that claim deserves a test in your own environment.
Kong's alliance work is still ahead. The company says it will build solutions with ServiceNow, take them through certification, and create joint offerings, and Drag's post promises more detail over the coming months. The enforcement Kong describes today applies to AI traffic your teams already route through Kong.
Several Members Describe Controls That Act While an Agent Runs
At least five founding members describe a control that works during an agent's run, and ServiceNow has one of its own:
- ServiceNow AI Gateway, announced at Knowledge 2026 in May, applies real-time controls to customer Model Context Protocol transactions.
- Kong AI Gateway enforces policy on model, tool, and agent-to-agent traffic.
- Cisco AI Defense Runtime Protection applies guardrails to running applications and agents.
- NVIDIA OpenShell enforces agent boundaries at runtime, according to Pat Lee, NVIDIA's vice president of strategic partnerships.
- Okta pairs its kill switch with ServiceNow's containment decisions to block new token requests from an agent acting beyond its scope.
- WitnessAI routes prompts based on the intent behind them to keep usage within policy and budget.
Some of these overlap. ServiceNow and Kong both run a gateway that can sit in front of tool calls made over Model Context Protocol, and Cisco and Kong both screen prompts and responses. Each company describes its own component in its own announcement. A customer running more than one will need a diagram that traces a single agent request through the stack and shows which gateway makes the allow-or-block call.
Cisco remains the source of the security finding, and ServiceNow holds the ticket and the owner.
Overlap at launch is normal for an alliance this broad, and two controls on a sensitive path can be a deliberate choice for defense in depth. The cost shows up in operations, because two gateways on one route means two policy stores to keep in sync and two sets of logs to reconcile after an incident.
What Happens Next
Cisco's post says alliance solutions stay compatible with ServiceNow releases, which addresses a familiar problem for any team that has seen an integration break after a platform upgrade. Three questions will shape how much of this customers wire up themselves:
- Will ServiceNow publish the shared standards for agent identity and guardrails so customers can compare partners against them?
- How often will certified solutions be retested as ServiceNow and its partners ship new releases?
- When two certified partners cover the same route, will ServiceNow recommend which one enforces and which one observes?
Before you expand a Control Tower contract or add a partner gateway, ask ServiceNow and each runtime vendor for one shared diagram of a single agent request to a model and a tool. It should show which component decides allow or block and where the evidence lands. If two vendors claim the same step, choose the one that enforces, and write that choice into the contract.
Cisco. "Cisco AI Defense Joins ServiceNow's Control Tower Alliance." Cisco Blogs, 28 Oct. 2026, cisco.com.
Kong. "Kong Joins ServiceNow Control Tower Alliance Aimed At Helping Enterprises Safely Adopt AI." Kong Press Room, 28 Oct. 2026, konghq.com.
Kong. "The Control Tower Reaches into the Runtime: Announcing the ServiceNow and Kong Partnership." Kong Blog, 28 Oct. 2026, konghq.com.
ServiceNow. "ServiceNow Launches Control Tower Alliance to Help Enterprises Adopt AI Safely at Scale." Business Wire, 28 Oct. 2026, servicenow.com.
ServiceNow. "ServiceNow Expands AI Control Tower to Discover, Observe, Govern, Secure, and Measure AI Deployed Across Any System in the Enterprise." ServiceNow Newsroom, 5 May 2026, servicenow.com.
