Describe the runbook. Mira builds it.
Natural-language IT automation. Durable, approval-gated workflows you author in plain English, and Mira runs them across your stack.
What it is, and what it does
Mira Runbooks lets your team declare AI-callable procedures in natural language, for example: "when a user asks to be added to a group, look up the group, ask their manager to approve, then add them and confirm." Mira generates the workflow, shows it to you on a visual canvas to refine, and deploys it. From then on, every active runbook is a tool Mira can call. When a chat, voice, or Teams request matches, Mira fills in the inputs and runs it, pausing at any approval gate you set.
How it works
Describe it
Say what you want in plain English. A multi-agent pipeline drafts the workflow.
Review it
See the generated steps on a visual canvas (actions, conditions, approvals, parallel branches) and refine.
Deploy it
The runbook becomes a live, typed, durable procedure Mira can invoke.
Mira runs it
When intent matches, Mira executes, stopping at your approval gates (Teams card, email, or in-app) and logging every step.
Author once. Mira runs it on request.
Representative interface. Sample data shown for illustration.
- Name
- Add user to a group and notify
- Description
- Add a user to a group after manager approval, then notify the user.
- LLM tool name
- add_user_to_group_and_notify
- Inputs
- user_emailstring, requiredgroup_namestring, requiredAdd input
- 09:41:02Run started(run_8f2ac1)
- 09:41:03Step suspended(approval_required)
- 09:43:18Approval decided(approved)
- 09:43:19Step started(add to group)
- 09:43:24Confirm to requester(core.send_email)
- 09:43:25Completed(run_8f2ac1)
A real runbook: "Add user to a group and notify"
Take one of the most common IT requests, adding a user to a group. The runbook has four steps: a trigger, a manager approval gate, the action that adds the user to the group, and a confirmation back to the requester. Mira exposes it as a callable tool (add_user_to_group_and_notify) with two inputs, the user's email and the group name. When someone asks in chat, Mira fills the inputs, routes the approval to the manager, adds the user on approval, and confirms. What used to sit in a ticket queue for days completes in a single flow.
Key capabilities
Natural-language authoring
Describe it; the canvas is the artifact, not the starting point.
Visual canvas builder
Drag-and-drop refinement when you want it.
Seeded playbook library
Start from proven runbooks, never a blank canvas.
Approval gates
Durable human-in-the-loop checkpoints via Teams card, email token, or in-app, with the decision recorded.
Runs history and replay
Per-step trace and audit on every execution.
Connect any system
Bring your own APIs and MCP tools into the action catalog, with governance metadata. Actions span your stack: Microsoft, Datadog, Entra, infrastructure, and Mira-native steps.
Why it is different
Legacy and point tools make you build automations node by node on a canvas, or bolt a planner onto a ticketing system. Mira leads with natural language (the canvas is the review artifact, not the authoring burden), and it is ITSM-native: runbooks read your incident and CMDB context and write the outcome back. An IT director who has lived through form-fills will choose "describe the runbook to Mira" over "drag twelve nodes onto a canvas."
See Mira Runbooks in your environment.
Book a walkthrough and watch Mira build a runbook and run it on your stack.