Is it possible to retain the full conversational context metadata when exporting AI bot interactions via the bulk export API for legal discovery purposes? We are observing a significant data gap in our chain of custody records. The environment is Genesys Cloud Europe/London, version 2024.3.0. When we trigger a bulk export job for digital channel transcripts involving the Pure Cloud AI bot, the resulting JSON files in our S3 bucket lack the specific intent confidence scores and the node traversal path taken by the bot. This data is critical for our legal team to reconstruct the exact decision-making logic of the AI during disputed interactions. The export configuration includes all standard transcript fields, yet the custom attributes defined in the bot flow designer, specifically the ‘intent_confidence’ and ‘dialog_node_id’ variables, are completely absent from the payload. We have verified that these variables are populated correctly during the live session by checking the real-time debug logs in the Architect flow. The issue appears isolated to the post-processing export phase. The S3 integration is functioning correctly for human agent recordings, where all metadata is preserved. However, for bot interactions, the export seems to flatten the data structure, removing the hierarchical context required for our legal hold compliance. We need to ensure that every step of the bot’s decision process is captured and immutable for discovery requests. Has anyone successfully configured the bulk export API to include these dynamic bot variables? We are considering using a custom Data Action to append this metadata before the S3 upload, but we want to avoid adding latency to the export job if a native configuration option exists. The current workaround involves manually querying the Interaction API for each specific transcript, which is not scalable for large volume discovery requests. Any guidance on preserving this metadata in the bulk export format would be greatly appreciated.