Ved3205/ai-meeting-intelligence ? reverse-engineered prompt

Build me an AI meeting intelligence app in Python that turns meeting recordings into a searchable knowledge base.

I want to be able to upload a meeting video or audio file, have it extract the audio, transcribe it, split the transcript into useful chunks, create embeddings, and store everything in Qdrant with meeting details and timestamps. Then I should be able to ask natural questions like what decisions were made, who owns an action item, or what was discussed about a topic, and get a grounded answer with the most relevant source chunks.

Please also include the basic meeting indexing flow, a query flow, and a simple summary flow since those are already part of the project direction. Keep it production oriented and organized so it’s easy to extend later with things like a proper web UI, user accounts, and more AI features. Use the existing README as the guide, and look up current docs online if you need to fill in any gaps.