Record calls and surface answers from your files

A Mac tool can transcribe live meetings and show sourced answer suggestions from uploaded documents as questions come up.

Done with Claude· Intermediate

What they did
They loaded PDF, text, and markdown files into Parrot’s knowledge base, where they were indexed on the Mac. During a call, Parrot captured system audio and microphone input without joining the meeting, then used local Whisper transcription. An assistant model such as Claude or Ollama watched the transcript and popped up answer cards tied back to the source document; the user could stop it at any time.
How it went
The site says cloud-based answer cards cost about $0.07 per hour of calls, while a fully local setup costs $0.00. It does not report accuracy, and the cards only work when a model is connected. In cloud mode, the transcript text is sent to the provider the user chose.
Tips
This requires macOS 14 or newer on Apple Silicon. If you want the most private setup, use the fully local path with Ollama and keep in mind that cloud models need your own API keys and send transcript text outward.

Try it yourself with Claude Code

Help me set up a meeting helper that listens to a live call from [audio source], transcribes it, and when it hears a question, searches the files in [folder path] and suggests a brief answer with the document name shown. Before any irreversible step such as installing, sending data, or connecting an API key, stop and ask me to approve it. I want to test it on a sample recording and see the questions, answer suggestions, and timestamps. Keep the workflow as local as possible and explain any data that would leave my machine.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on openparrot.app ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.