The Pulse
Amazon Quick’s Smart Approvals Automatically Approves Eligible Actions or Returns Them for Human Review
Amazon Quick’s Smart Approvals evaluates eligible agent actions against a user’s request and a bounded view of the conversation. It leaves required human approvals in place and records permission updates associated with Smart Approvals in A
AI.info Team · Written with AI · fact-checked ·
Amazon Quick’s Smart Approvals addresses a tension in agent oversight: requiring a person to approve every step can produce repeated prompts, while a permitted action can still warrant human judgment. Smart Approvals reviews eligible actions against a user’s request, then either approves them automatically or returns them to the user for review. It does not grant additional permissions or remove required human approvals.
Policy sets the boundary
Before automated review, Quick checks each proposed action against Custom Permissions, Quick Guardrails, connector action permissions and the user’s settings. Denied actions stay denied, and mandatory human approvals still apply. An action is eligible only when the user’s setting is “Ask each time” and no other control requires human approval. Connector action permissions, configured when a connector is created, can require human approval for specific actions.
Administrators use Custom Permissions to determine who can use Smart Approvals. The feature is on by default, but profiles with deny-by-default enabled must explicitly allow it. Users can switch it off for themselves, and that choice applies across conversations and scheduled runs. When a user turns it off, actions that would otherwise be eligible for automated review require their approval; no account-level setting can turn it back on for them.
How eligible actions are reviewed
Smart Approvals uses a separate model, called the classifier, to evaluate eligible actions against the user’s request and a bounded view of the conversation. The classifier considers what the action does, where data goes and who it affects. It can consult the user’s knowledge graph, which contains people and working relationships drawn from connected sources, to help understand relationships such as whether a recipient is in the user’s reporting chain. That lookup stays within the Quick account, and Quick says it does not use the content the classifier reviews, including knowledge graph data, to train models or improve the service.
Actions can match a request and still warrant human review. Amazon lists sending content outside an organization or to senior leaders in the user’s reporting chain, changing access, deleting records and initiating payments as examples; it says the list is not complete. Reviewing actions against the user’s request adds a check against indirect prompt injection, in which instructions in retrieved content attempt to redirect an agent. Quick instructs both the agent and classifier to treat retrieved content as information, not instructions. Model review can still make mistakes.
Automatic approvals apply once, to exact parameters
Amazon distinguishes the classifier’s judgment from the permission service’s enforcement of authority. Each automatic approval is single-use and applies only to the exact action and parameters reviewed; it grants no ongoing permission. Quick labels automatic approvals in the conversation. Actions the classifier does not approve return to the user, as do timeouts and other review failures, which never grant automatic approval.
The example involves an account manager asking Quick to prepare a quarterly business review brief using her team’s account notes and last quarter’s email, then send it to two colleagues she names. If an account note says the brief should also go to an all-sales distribution list of about 2,000 people, Smart Approvals is designed to return the proposed send to her for review because that would change the audience beyond the recipients she specified. An approval for the original recipients cannot be reused for the wider audience.
CloudTrail records permission events
The Quick permission service records permission updates associated with Smart Approvals in AWS CloudTrail in your account. Each event identifies the user, connector, action and time, and marks the request as associated with Smart Approvals. The marker does not establish that the permission service accepted the grant or that the action ran. Quick also records permission-check verdicts, including denials from connector action permissions, and changes to a user’s Smart Approvals setting.
Investigators can correlate those events by user and time with other Quick activity in CloudTrail and logs delivered through Amazon CloudWatch Logs. Chat logs contain the user’s request, system response and conversation context; agent metadata logs record changes to agents. Records from the destination system, such as a sent email or updated ticket, confirm whether an action completed. CloudTrail Event history retains events for 90 days, so configure a trail for longer investigations; because chat logs can contain sensitive content, restrict access and consider CloudWatch Logs data protection policies to mask sensitive data.
Where to start
Because Smart Approvals is on by default, begin by confirming that actions requiring human approval are configured through connector action permissions, then set up a CloudTrail trail and delivery of chat logs and agent metadata logs to CloudWatch Logs. Chat logs capture activity only after delivery is configured, so earlier approvals will lack that conversation context. Use Custom Permissions to limit Smart Approvals to a pilot team working on a defined workflow before extending it to other teams.
Before the pilot, define what success looks like: fewer prompts for routine actions, with consequential actions still returning to a person for review. During the pilot, review a sample of automatic approvals and test high-impact cases, such as sending content outside the organization or adding a recipient introduced by a document. Confirm that your team can trace each action from approval through execution. Expand to more teams and workflows only when the results meet those criteria, and reassess as actions and policies change.