Place a reviewed draft between AI interpretation and a consequential cloud-flow action.
Workflow & automation · Documented · Documentation checked 2026-10-04
Cloud flows, desktop flows, AI Builder prompts, and Power Platform environments.
Work or school account, licenses, connectors, regional AI capacity, and environment policy affect availability.
An illustrative recipe. This website does not run the automation.
The goal and the right kind of system
Place a reviewed draft between AI interpretation and a consequential cloud-flow action.
A fixed cloud flow fits because input, prompt, validation, approval, and action are known. A general agent is unnecessary.
Inputs and connected services
Inputs: work / school identity, test environment, sample question / agenda, AI Builder access, and authorized reviewer.
Services: Power Automate, AI Builder, approval channel, and a narrowly scoped test connector.
The steps
An illustrative sequence to adapt and test before using real data or enabling ongoing work.
- Create an instant sample flow with controlled inputs.
- Add Run a prompt and inspect the draft output.
- Validate the draft and request approval for exact destination and content.
- Branch on the real approval result and test a sandbox action once.
Outputs and human checkpoints
Outputs: draft, approval decision, and inspected action result or visible exception.
Human checkpoints: verify source support and the exact recipient / record / content before approval.
Exceptions, permissions, and stopping conditions
Permission boundary: no model-authored approval state. Reject, absent decision, or changed content stops the action route.
Stop on invalid output, policy block, missing capacity, denied approval, or uncertain completion. Reconcile the destination before retrying.
Adaptable prompt
Create a draft from the approved agenda and sample question. Cite the relevant passage and identify missing details. No action is authorized by your output. A separate flow validation and human approval decides any external operation.
Related reading
Cloud flows, desktop flows, and connectors
Choose an event, button, or schedule trigger and identify where each action actually runs.
AI Builder: Run a prompt
Map controlled inputs into an AI Builder prompt and inspect its output before taking action.
Approval flows and human decisions
Route a proposed decision to an authorized reviewer and branch on the actual approval outcome.
Environments, data policies, and governance
Check which connectors may exchange information before activating an AI-enabled flow.
Designing an AI workflow
Make a useful task repeatable with clear inputs, steps, outputs, exceptions, and human checkpoints.
Permissions and meaningful approvals
Decide what a system may read, what it may change, and when someone needs to authorize a concrete action.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Types of Power Automate flows · Microsoft
Run a custom prompt in a flow · Microsoft
Create and test approvals · Microsoft
Power Platform data policies · Microsoft
Desktop flow governance · Microsoft