Recurring reporting
Collect approved context, prepare the analysis, and notify the report owner on the chosen cadence.
Schedule agents, react to system events, and notify the right owners while permissions, confirmations, approvals, and autonomy controls stay part of the workflow.
A Routine can use agent reasoning while staying inside configured access and decision boundaries.
Collect approved context, prepare the analysis, and notify the report owner on the chosen cadence.
Start relevant agent work when a candidate, employee, skill, review, or another supported object changes.
Notify the right person when work is relevant to them or when a confirmation, approval, or judgment call is needed.
Routines give repeated People work a trigger, approved context, an agent, a clear owner, and the right human review points.
Run agent work on a schedule
React to object changes and incoming webhooks
Send relevant updates through Clous, email, Slack, or Microsoft Teams
Require confirmation or approval before record changes by default
A trigger does not give an agent access to context the user, role, company, document, or integration cannot use.
Authorized retrieval may run directly; changes to records or objects require confirmation or approval by default.
More permissive behavior depends on configured autonomy controls and never bypasses visibility or decision boundaries.
Use the trigger that matches how the People workflow actually begins.
Run recurring reporting, analysis, checks, follow-up, or other agent work on a chosen cadence.
Start work when a Clous object changes or an approved incoming webhook signals a relevant event.
Let an authorized person launch the same governed sequence when judgment determines the right moment.
Move repeated coordination into the People operating system without removing accountable owners.
Consistent follow-through
Give recurring work a defined trigger, context, owner, notification path, and review boundary.
Reusable agent work
Use the same specialized agents and company context across scheduled, event-driven, and human-started work.
Selective autonomy
Choose where a Routine proceeds, asks for confirmation, requests cross-user approval, or escalates.
Start with a repeated workflow whose trigger, inputs, owner, and decision boundaries are already understood.
Define the schedule, object change, incoming webhook, or human action that should start the work.
Choose the data the agent may retrieve, the actions it may prepare, and where confirmation, approval, or escalation is required.
Inspect the output, corrections, and outcome before expanding the Routine to adjacent work.
Bring one recurring workflow. We will map its trigger, context, agent work, owner, and human-control points.