Home › Key concepts › Human approval
Key concept · Walk (intermediate) · the same on every platform
Human approval
A person says yes before anything that is costly or cannot be undone.
What it is
The system pauses and asks a named person to approve risky actions, such as payments above a limit, emails to customers, or deletions, showing exactly what will happen. The approval is recorded.
See it happen: Problem Lab: self-healing with a runbook and gates, not an agent in production (simulation), recorded from a real run.
Why it matters
It keeps people accountable for decisions that matter while the system does the legwork.
An example
A refund over ₹50,000 waits for a supervisor, who sees the claim, the amount and the reason, then approves.
The common mistake
Asking for approval on everything, so people click yes without reading.
Where you meet it on each platform
The idea stays the same; only the product name changes.
| Job | AWS | Azure | Google Cloud | Databricks | Open source |
|---|---|---|---|---|---|
| Agent runtimewhere a long-running agent keeps its state and waits safely | Bedrock AgentCore | Foundry Agent Service / Durable Functions | Vertex AI Agent Engine | Mosaic AI Agent Framework | DBOS or Temporal + your loop |
Related ideas
Where it shows up
Part of the Walk level on these paths: AI Engineering (tools and agents).