I inherited this Genesys Cloud org about six months ago and I am currently struggling with a very specific analytics issue. We are attempting to pull intent recognition performance data for our Dialog Engine Bot Flows using the Python SDK. However, the get_analytics_botflows_sessions function appears to be returning inconsistent execution IDs. When I cross-reference the IDs against the conversation detail records, about thirty percent of them are missing or malformed. Has anyone else encountered this bug in the SDK, or is there a specific way I should be formatting the query to ensure I get valid execution IDs for every session?
Good afternoon. I manage a large fleet of on-prem Edge servers and I frequently interact with the internal media services that generate these IDs. I can confirm that the execution IDs are generated within the bot engine microservice, not the Python SDK itself. The ‘bug’ you are seeing is likely due to the asynchronous nature of the bot analytics data lake. If you query the API too soon after the session ends, the execution ID might not have propagated to the analytics index yet. You must implement a polling delay or use a ‘Retry-After’ logic in your Python script to ensure the data is fully indexed before you attempt to retrieve the IDs.
I agree with the previous assessment from a performance dashboard perspective. If you look at the raw JSON in the Performance Workspace, you will see that the execution IDs are often added as a secondary metadata tag. In my experience, if the bot flow encounters an error and fails over to a human agent, the execution ID for the bot session is sometimes overwritten by the primary conversation ID. This makes it look like the ID is malformed when it is actually just a different object type. You should check the flowId attribute in your Python script to differentiate between a successful bot turn and a failure handoff.
Actually, this is a known issue that I have reported on the Genesys GitHub repository for the Python SDK. The problem is in the deserialization of the BotFlowSession object. The SDK expects a specific GUID format, but the bot engine sometimes returns a slightly different string structure for legacy Dialog Engine flows. I have submitted a pull request that adds a more flexible regex for ID validation. Until the fix is merged into the official release, you can work around this by accessing the raw dictionary response via response.to_dict() instead of relying on the typed object properties. It is a bit messy, but it will let you capture those malformed IDs correctly!