Hey guys, this is my first dev job and they told me to write a Python script that downloads all the chat history from our new Dialog Engine bot. My boss wants a report on what questions the bot gets wrong. I found the /api/v2/analytics/bots/aggregates/query endpoint which gives me the high level numbers, but I actually need the step-by-step transcript of what the bot said versus what the customer typed. I tried looping through every single conversation ID and pulling the recording, but my script crashes after about three minutes. Is there a specific API for getting bot conversation turns without looping through recordings?
I recently transitioned from PureConnect, where bot interaction details were stored in a relational database. In Genesys Cloud, you should not loop through individual recordings for analytics. The correct approach is to utilize the Conversation Details API.
Query /api/v2/analytics/conversations/details/query with a segment filter for segmentType: bot. This returns the metadata for the bot session.
However, to extract the literal conversational turns, you must configure a Dialog Engine Transcript action within the bot flow to save the transcript to a data table or external system during the interaction.
The previous reply is correct regarding the architecture, but let me warn you before you build this Python script. If you attempt to query the Conversation Details API synchronously for thousands of bot interactions, you will trigger the unpublishable API rate limits. The analytics queries are strictly governed by a complex concurrency limit.
If your script spins up multiple threads to query conversation details, the gateway will instantly throw a 429 error and temporarily ban your OAuth token. You must use the Asynchronous Analytics Export API (/api/v2/analytics/conversations/details/jobs) to request the data in bulk, wait for the job to finish, and download the single JSON file.