What ServiceNow Agentic AI includes
ServiceNow's AI Agents combine agents, agentic workflows, tools, data, and platform controls. In current documentation, AI Agent Studio lets teams create, manage, or test agents and workflows. An agent is configured with an identity, descriptions and instructions, tools, and security settings. Its tools can be script tools, subflows, or knowledge retrievers.
| Capability | What it does | What a team still decides |
|---|---|---|
| AI Agent Studio | Creates, manages, and tests agents and agentic workflows | Objective, instructions, tools, data access, and rollout criteria |
| AI Agent Orchestrator | Coordinates multiple agents toward a workflow goal | Which work should be delegated and how failures or missing context are handled |
| AI Agent Analytics | Surfaces agent and workflow usage and performance information | Which metric is meaningful and what triggers an intervention |
| AI Control Tower | Provides central AI discovery, security, governance, observability, and value measurement | Ownership, policy, risk thresholds, and remediation process |
Agentic work is still a configured workflow with permissions, not a free pass to act on every request.
The panel shows the steps an AI-agent workflow takes for a case. Source: ServiceNow AI Agents.
Start with a job that has a clear boundary
Choose a job with a clear owner, permitted data, and a measurable end state. “Resolve every incident” is too broad. “Collect required facts for a password-reset incident, create a draft update, and route exceptions to the on-call engineer” creates boundaries someone can test.
In AI Agent Studio, ServiceNow offers both manual execution tests and automated agentic evaluations. Use manual tests for an expected path, then use broader evaluations to find repeatable failure patterns. Also test access controls: ServiceNow documents a specific test mode for security controls. A tool being available in configuration is not evidence that every requester should be allowed to use it.
Review these cases before deployment:
- The expected task with current records and knowledge.
- A similar request where the record belongs to another user or team.
- A request with missing data, stale knowledge, or conflicting instructions.
- A tool failure or a request outside the agent's approved domain.
- A consequential action, such as closing an incident or changing access, that must remain held for a human.
For every case, inspect the execution log and verify the final record or connected-system result. If the agent says it updated an incident, look at the incident; if it sends a customer update, check the right channel received the intended message.
Where eesel CLI fits for support work
ServiceNow Agentic AI and eesel CLI address different systems. ServiceNow AI Agent Studio configures ServiceNow agents. eesel CLI operates a ready-to-work eesel helpdesk teammate. The CLI and dashboard point to the same eesel workspace and teammate, so a support owner can stay in the dashboard while an engineer, script, Claude Code, Codex, or Cursor consumes the CLI's JSON output.
For example, a support lead may want an eesel teammate to give a safe update after a ServiceNow incident is created. A coding agent can inspect whether that teammate has the approved status-page guidance and an instruction to hand off major incidents instead of inventing an ETA. It should report the gap first, not connect ServiceNow, change an action, or enable automation as a side effect.
With Node.js 18.17 or newer, start with an explicit target rather than a saved default:
npx @eesel/cli login
npx @eesel/cli whoami
TARGET_EESEL_AGENT="paste-agent-id-or-name-here"
npx @eesel/cli status --agent "$TARGET_EESEL_AGENT"
npx @eesel/cli integrations --agent "$TARGET_EESEL_AGENT"
npx @eesel/cli instructions --agent "$TARGET_EESEL_AGENT"
After the support owner approves a changed incident instruction, use --dry-run on the relevant supported write before sending it. Before a test chat, review the selected teammate's actions and permissions. Disable or contain actions outside the test scope; a held-action queue does not protect against an action that can run without a hold.
Then use a fictional incident follow-up to test the policy boundary, not ServiceNow itself:
npx @eesel/cli new --name "incident-status follow-up" --agent "$TARGET_EESEL_AGENT"
npx @eesel/cli chat "A customer asks for an ETA during a major incident. Use the approved status guidance, avoid inventing a restoration time, and state when the on-call team must respond." --agent "$TARGET_EESEL_AGENT"
npx @eesel/cli approvals --agent "$TARGET_EESEL_AGENT"
npx @eesel/cli activity --agent "$TARGET_EESEL_AGENT"
Review the reply, held actions, and activity. Then verify the real ticket, status page, or ServiceNow record after a controlled rollout. CLI chat is billed work; its dashboard-management commands are free. JSON output makes a coding agent's report inspectable, but it does not replace a support owner's approval or an external outcome check.
Choose the system that owns the work
Use ServiceNow Agentic AI when the goal requires an agentic workflow inside ServiceNow and the team will own its roles, records, tools, testing, governance, and operation. Use eesel when the immediate job is a helpdesk teammate that works with approved company knowledge and support workflows. The CLI provides a repeatable terminal route into that same teammate rather than a separate support configuration.
Test incident follow-up with eesel
Try eesel if you want to test an approved helpdesk teammate response before expanding incident automation. Start with the correct knowledge and a narrow case set, contain unrelated actions, and verify the real support or ServiceNow result after rollout. Check eesel pricing separately from ServiceNow licensing because they are different products and units of work.
Frequently asked questions
What is ServiceNow Agentic AI?
ServiceNow AI agents are configured agents that use instructions, tools, record context, and searchable content to work toward a goal. In AI Agent Studio, teams can create, manage, and test agents and agentic workflows.
What is AI Agent Studio used for?
AI Agent Studio is ServiceNow's place to create, manage, and test AI agents and agentic workflows. It includes ready-made options, activity logs, manual testing, and automated agentic evaluations, subject to the instance's installed products and access.
What does ServiceNow AI Agent Orchestrator do?
The AI Agent Orchestrator coordinates collaboration among AI agents for an agentic workflow. It does not remove the need to define a clear objective, map the permitted tools, set access controls, and test what happens when an agent cannot complete a task.
What should be tested before ServiceNow agents go live?
Test normal tasks, missing or conflicting records, authorization changes, ambiguous requests, failed tools, and actions that need a human hold. Review the execution log and verify the actual record or external system outcome, not only an agent response.
How does ServiceNow control agent access?
ServiceNow uses configured tools, security controls, roles, and data access settings. An agent needs access to the records and tools it is meant to use. Apply least privilege and confirm the controls against representative requests before deployment.
Does eesel CLI operate ServiceNow AI Agent Studio?
No. eesel CLI operates an eesel teammate, not ServiceNow Agent Studio or its workflows. It lets people, scripts, and coding agents inspect and test the same eesel teammate and workspace available in the dashboard.
Article by
Alicia Kirana Utomo
Kira is a writer at eesel AI with a Computer Science background and over a year of hands-on experience evaluating AI-powered customer service tools. She focuses on breaking down how helpdesk platforms and AI agents actually work so that support teams can make better buying decisions.
