ServiceNow AI Footnotes
What ServiceNow's AI features do with your data, as ServiceNow publishes it. Checked October 31, 2026. ServiceNow in the Directory.
Agent access
MCP Server Console lets admins create MCP servers on an instance, including a preconfigured Quickstart Server. It ships through the ServiceNow Store with activation of a Now Assist application, and the product tiers page places inbound MCP Server Console in the Prime tier. ServiceNow calls it a new application in the Brazil release and does not state a release stage. source, source 2, source 3 · checked October 31, 2026
Now Assist skills
ServiceNow AI Platform (ServiceNow Otto applications; Foundation, Advanced and Prime tiers)
| On by default? | Since Zurich Patch 4, designated generative AI skills turn on automatically when an AI product is installed. On upgrade, an asset that was never turned on is turned on unless an admin turned it off or changed its roles. source, source 2 · checked October 31, 2026 |
|---|---|
| How an admin turns it off | Skills are activated one by one in All > AI Admin Hub > Skills, with the sn_generative_ai.nsa_admin role. The Brazil page gives activation steps only. source · checked October 31, 2026 |
| Trains on customer data? | Yes unless opted out: data shared through the customer data-sharing program is used to improve Now LLMs, and customers can opt out. source, source 2 · checked October 31, 2026 |
| Retention | For processing in ServiceNow's compute hubs (Now LLM Service, and Azure OpenAI hosted in those hubs), the data used to generate a response is deleted once the response is generated, and inputs and outputs are not cached or stored there. The page does not address Google Gemini or AWS Claude, the third-party default providers since July 2026. source, source 2 · checked October 31, 2026 |
| Where data is processed | Admins choose regional data routing, which keeps LLM requests in datacenters within the region, or global routing, in AI Control Tower's AI model providers settings. source · checked October 31, 2026 |
| Model providers | Now LLM Service (ServiceNow-hosted models, adding Gemma 4 26B-A4B in October 2026) and the integrated third-party providers Google Gemini, AWS Claude and Azure OpenAI. source, source 2 · checked October 31, 2026 |
| Price | Metered in assists against purchased entitlements; each skill action consumes assists at that skill's assist ratio. source · checked October 31, 2026 |
AI Agents
AI Agent Studio on the ServiceNow AI Platform (Foundation, Advanced and Prime tiers)
| On by default? | When an AI product is installed, designated AI agents and agentic workflows are turned on automatically. source · checked October 31, 2026 |
|---|---|
| How an admin turns it off | Admins with the sn_aia.admin role deactivate an AI agent or agentic workflow in All > AI Agent Studio > Agentic solutions, then select Deactivate. source · checked October 31, 2026 |
| Trains on customer data? | Yes unless opted out: the customer data sharing program provides data to improve ServiceNow AI products; a data steward opts out per instance. source · checked October 31, 2026 |
| Retention | Not published as of October 31, 2026 |
| Where data is processed | Now Assist data is processed in ServiceNow datacenters and can burst to Microsoft Azure datacenters at high traffic. Instances are opted in by default; a data steward can opt out. source, source 2 · checked October 31, 2026 |
| Model providers | Third-party models are the default for skills and agents; customers can select the Now LLM Service instead. source · checked October 31, 2026 |
| Price | Included by product tier: out-of-the-box agents at Foundation and Advanced; building new agents from natural-language instructions requires Prime. source · checked October 31, 2026 |
AI Agent Orchestrator
AI Agent Studio agentic workflows on the ServiceNow AI Platform
| On by default? | Not published as of October 31, 2026 |
|---|---|
| How an admin turns it off | Deactivating an agentic workflow in AI Agent Studio stops it from running. source · checked October 31, 2026 |
| Trains on customer data? | Data sharing is never enabled on GCC or self-hosted instances; commercial instances share unless a data steward opts out. source · checked October 31, 2026 |
| Retention | Not published as of October 31, 2026 |
| Where data is processed | Not published as of October 31, 2026 |
| Model providers | Not published as of October 31, 2026 |
| Price | Assist consumption is reported per asset type: skills, agents and agentic workflows. source · checked October 31, 2026 |
Connections to outside MCP servers
AI agents built in AI Agent Studio can use tools from external MCP servers through the Model Context Protocol Client. A server is added in AI Agent Studio with OAuth 2.1, an API key, or a connection and credential alias. ServiceNow does not state a release stage. source, source 2 · checked October 31, 2026
Compliance
EU AI Act
ServiceNow signed the EU General-Purpose AI Code of Practice under the EU AI Act (blog post, September 18, 2025). source · checked October 31, 2026
Not published by ServiceNow as of October 31, 2026
- List price of an assist or of Now Assist entitlements
- Assist ratio for each skill, agent and agentic workflow (shown only inside an instance)
- Retention period for AI agent execution logs and long-term memory
- Retention by Google Gemini and AWS Claude when they serve Now Assist requests
- Brazil-release steps to deactivate a generic Now Assist skill
- Which AI products and features are unavailable in the FedRAMP data centers (published only in a Now Support article)
- HIPAA BAA scope for Now Assist, in the Now Assist documentation
- Release stage of MCP Server Console and the Model Context Protocol Client
Changes
No changes recorded since the first check on October 31, 2026.
How this page is made, and how to send a correction: method.