May 20
Better AI post-processing
As a user, I want to be able to post-process my meeting recordings in chunks or sequential steps using large language models.Currently, the post-processing for an entire meeting is performed in a single pass. Because LLMs often have limitations regarding maximum output tokens, they occasionally truncate the results or provide placeholder summaries like "and then the meeting continued" for later segments. I would like to propose a feature that implements a processing loop or "chunking" workflow. This would allow the system to process lengthy meetings in stages, ensuring that no data is dropped and the model can provide high-quality, comprehensive output for the entire duration of the recording.
Pending