Route only valid approved records to consequential actions and stop safely on failures.
Workflow & automation · Documented · Documentation checked 2026-10-04
Make scenarios and Make AI Agents (New); connected model providers have their own terms.
Scenario limits vary by plan. AI Agents (New) is in open beta; custom AI providers require a paid plan.
The worked example is illustrative and has not been tested in a signed-in product account.
What it does and how it connects
Route only valid approved records to consequential actions and stop safely on failures.
Filters pass bundles that meet defined conditions; rejected bundles stop on that route. Error handlers address module failures. A review gate can be implemented as a validated approved state before a write; asking an agent to review itself does not enforce that gate.
Walkthrough
Use the product named in the access panel. The steps below combine documented controls with the editorial example.
- Add a filter on the documented connection between modules.
- Define exact valid-input and approved-state conditions from controlled fields.
- Route incomplete or rejected records to a private review output.
- Configure documented error handling and test a failure before enabling live actions.
Worked example
An AI classifies workshop requests, but only a person can set the separate approval field used by the send route. The model’s suggested approval must never be mapped into that authoritative field.
Illustrative gate
Required: valid request_id, verified recipient, human_approved = true
Authority: human_approved comes from the review record, never model output
Allowed action: send the reviewed content once
Rejected / missing approval: stop and retain draft
Error: record exception; no blind resend
Inspect and verify the result
Test approved, rejected, and absent-approval cases. Inspect the actual downstream operation count and verify no action occurred on the blocked route. Simulate an unavailable service and inspect the exception behavior.
Limits and troubleshooting
A filter can inspect the wrong field or accept a loosely typed value. Validate the authoritative review record and exact content. Retrying a write after an uncertain response can duplicate it; reconcile the destination before retrying.
Data, permissions, and actions
Approval records and error logs may contain private information. Restrict reviewer access and connector credentials. Keep the irreversible operation after the gate and make the stop route visible to the operator.
Related reading
AI Agents (New): tools and configuration
Configure a beta agent with a bounded goal, named tools, and explicit execution limits.
Fixed AI triage with a review filter
Classify synthetic requests while keeping authoritative review separate from AI output.
A bounded beta knowledge agent
Use the beta agent to retrieve approved workshop evidence and produce a validated response draft.
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.
Filtering bundles · Make
Error handling · Make