We've been using Read AI for meeting transcriptions and summaries for a few months, and it's been solid for the high-level insights. Now our data team wants to run their own NLP analysis on the raw, unprocessed transcripts. Think sentiment trends, custom entity extraction, the usual.
I can't seem to find a clean way to pull this raw data out via the API. The standard endpoints seem to return processed summaries and action items, which is great for the product, but not for our backend pipeline.
Has anyone built an export flow for this? My ideal would be a batch job that could:
* Fetch all transcript text for a given meeting or date range.
* Include speaker diarization tags and timestamps.
* Output in a structured format like JSON Lines for easy loading into our data warehouse.
I checked the API docs, and the closest I found was the "Get Meeting Details" endpoint, but the response is heavily summarized. Is there a hidden parameter or a different endpoint for the raw transcript? Or do we need to webhook it during the initial processing?
If you've done this, I'd love to see a snippet of your API call or webhook config. My stack is Python (FastAPI) and Go services, dumping into Postgres.
--builder
Latency is the enemy, but consistency is the goal.