Architect Flow Execution Timeout in Multi-Org Premium App Integration

Architect & Flows
1

Dealing with a very strange bug here with the Architect flow execution when triggered from our Genesys Cloud Premium App integration. The issue manifests as a 504 Gateway Timeout when the flow attempts to make a series of HTTP requests to our external backend service within a single flow action block. This is happening specifically in our multi-tenant environment where each client organization has distinct OAuth scopes and rate limit configurations.

Here are the steps to reproduce the issue:

  1. Deploy a Premium AppFoundry integration that registers a custom webhook endpoint for event processing.
  2. Configure an Architect flow with a single “Make HTTP Request” action that calls our backend service at https://api.ourpartner.com/v2/process-data.
  3. Set the timeout value in the HTTP action to 30 seconds (the maximum allowed by the platform).
  4. Trigger the flow from a simulated inbound call or via the API using the /api/v2/architect/flows/{flowId}/execute endpoint.
  5. Observe the flow execution logs in the Genesys Cloud Admin portal.

The flow consistently times out at the HTTP request step, even though our backend service responds within 2-3 seconds for the same payload when tested directly via Postman. The error log shows:

Error: Request timed out after 30000ms
Endpoint: https://api.ourpartner.com/v2/process-data
Flow ID: 1a2b3c4d-5e6f-7g8h-9i0j-k1l2m3n4o5p6

We have verified that:

  • The external endpoint is accessible from the Genesys Cloud network (no firewall blocks).
  • The OAuth token used for authentication is valid and has the correct scopes.
  • The payload size is well within the 1MB limit.
  • The issue does not occur when the flow is executed in a sandbox organization with lower traffic.

Our hypothesis is that there might be a rate limiting or connection pooling issue at the platform level when multiple organizations are triggering flows simultaneously. The integration is designed to handle high-volume call routing, so this timeout is causing significant delays in our customers’ IVR experiences.

Has anyone encountered similar timeout issues with HTTP requests in Architect flows within a multi-org Premium App context? Are there any known workarounds or platform-level configurations that could help mitigate this? We are considering breaking down the HTTP requests into smaller, parallel actions, but that adds complexity to the flow design. Any insights or debugging tips would be greatly appreciated.