🤖 Autonomous IT Academy Recap: Your Top Questions on AI Agents & Otto for Incident Management

[aastha14]

Thank you to everyone who joined our Autonomous IT Academy session on Activating AI Skills and Agents for Incident Management! Here's a full roundup of what the community asked, organized by topic including documentation links for the questions that were answered live in the moment.

Licensing, SKUs & Packaging

Q: How do I know which AI agents/skills are included in each SKU tier (Foundation, Advanced, Prime)?
Each SKU tier maps to different levels of AI maturity and use-case adoption. Foundation covers a handful of skills, while Prime covers the full set. Connect with your account executive for tier specifics, or see the AI-Native SKU overview docs.

Q: What are the prerequisites for enablement, and is there a usage cost for Now Assist?
Customers should be on an AI-native SKU to start. Each AI agent workflow consumes "Assists" similar to tokens in the broader AI world.

Q: Is a Statement of Work (SOW) mandatory for an Otto implementation?
Not mandatory, but encouraged for a future-ready AI experience.

Q: Is there a way to see the cost/value Otto provides?
Reach out to your account team, assist consumption ties directly to skill/agent usage.

Q: Does the ITSM Prime license include AI Control Tower, Agentic Studio, and Custom Agents automatically or do these need separate installation?
AI Control Tower, Now Assist, Moveworks, and Workflow Data Fabric are bundled into every ITSM tier, including Foundation but custom AI skill and agent creation, along with the AI Agent Studio and AI Agents for ITSM, are exclusive to the Prime tier. So AI Control Tower itself isn't Prime-exclusive, but building your own custom agents is. If your instance shows fewer plugins than expected, that's likely a tier/entitlement gap worth confirming with your account team rather than a delay in plugin availability.

Otto Branding & Availability

Q: Is Otto tied to a specific release, or available in earlier releases?
Otto isn't tied to one release, it's a rebrand of the AI experience linked to plugin versions. Newer plugin versions are more likely to show Otto branding.

Q: What plugin needs to be updated to reflect the Otto branding?
Same plugin as before, "Now Assist for ITSM" (now shown as "ServiceNow Otto for ITSM").

Implementation, Testing & Access

Q: Can we enable Otto in non-production instances to test before go-live?
Yes.

Q: Can Partners get access to the L1SD specialist in demo instances?
Yes, Demo Hub narrative access for partners is coming soon.

Q: Will there be an API to connect to local/on-prem AI models?
Yes, see the ITSM MCP server docs for calling AI agents and tools.

Q: How do you invoke an agentic workflow from an existing workspace form?
Agents can be discovered and run directly from the Otto (Now Assist) panel inside a workspace, with the guide covering prerequisites and how to create and test them there — see How to run and test AI agents in workspaces and Virtual Agent.

AI Admin Center & AI Control Tower

Q: Is customer sentiment data on the incident record reportable?
Yes — via the AI Admin Center dashboard.

Q: Is the AI Admin Center available on Yokohama/Zurich, and does it need a separate plugin?
See Now Assist Center landing page docs for release-by-release availability.

Q: Does AI Admin Center monitor all workflows?
Yes.

Q: How do AI Admin Center and AI Control Tower (AICT) differ, given the feature overlap?
AI Admin Center is the operational, day-to-day view — showing usage, performance, and business-value analytics on its home page — while AICT's "Measure" capability is the governed, cross-asset value record used to report up to sponsors/leadership. In short: Admin Center = run it day-to-day; Control Tower = govern and report on it enterprise-wide. Reference: ServiceNow AI Agents — Build, Operations & Governance Guide.

Q: Can we debug agents and their behavior in AI Admin Center?
There's a documented step-by-step debug sequence covering which tables to check when an AI agent fails, referenced in the ServiceNow AI Agents Build, Operations & Governance Guide.

Incident Summarization & Categorization

Q: Is there a structured way to test whether a different LLM performs better for incident summarization?
No built-in comparison tool today best approach is comparing generated content and fulfiller acceptance rates.

Q: Will categorization work if we've fully customized our category/subcategory choices (still using OOB fields)?
Yes, it relies on OOB field names, not the specific choice values.

Q: For an OOB agent/workflow, are triggers pre-defined, or built from scratch?
Most agents ship with a base trigger already configured, but you can customize or add your own.

Q: Can incidents be converted into Knowledge Base articles?
Yes, this can be demoed.

Q: Can Now Assist help remove PHI/PII/CUI from records?
This is something AI Agents can be configured to do — ServiceNow's "AI Agents: Configure AI Agents" course on Now Learning walks through setting up a PII Removal Agent. There's also a dedicated ServiceNow Data Privacy product (Data Classification, Data Discovery, Data Anonymization pillars) for this at a broader instance level — worth exploring both.

Q: Where can I find a list of all available agents and their functionality?
See Now Assist Skills documentation.

---

Here is the post-webinar blog post! Activate AI Skills and Agents for Incident Manag... - ServiceNow Community