AWS
Invoke your own AWS Lambda functions from a workflow.
The AWS integration lets you run custom logic by invoking your AWS Lambda functions directly from a workflow — anything you can express as a Lambda becomes a step in a flow.
Setup
Connect an AWS integration on the Integrations page in the app, supplying the credentials and region for the account whose functions you want to invoke. The Invoke Lambda node references the connected integration for its credentials, so you select which integration to use on the node itself.
What you can do
- Invoke a function — call any Lambda in the connected account, optionally targeting a specific version or alias with a qualifier.
- Pass data in — send a JSON payload that can embed
$(name)variables, so the function receives values gathered earlier in the flow. - Use the result — capture fields from the returned JSON into variables by JSONPath and branch on or reuse them downstream.
- Fire and forget — invoke asynchronously when you don't need to wait for a result.
The Invoke Lambda node exposes a timeout exit for when the function has not answered within
the node's Timeout plus one second, or 28 seconds, whichever is shorter. Wire it so the flow
degrades gracefully. Your function keeps running after the node takes timeout: the node stops
waiting, but the invocation is not cancelled. See Error
Handling.
When the function fails
If your function throws, or runs past its own configured Lambda timeout, the node takes
fallback. The message it carries is the function's errorType: errorMessage, or errorType
alone when the function reported no message. What a Lambda timeout reads as depends on the
function's Lambda runtime, for example Sandbox.Timedout: .... No stack trace or other part of
the error payload is included, and the type and message are cut at 100 and 500 characters. Inside a Try / Catch, the message is in $(sys.error.message).