The basic idea came from liking the bot free experience of tools like Granola, but not liking that the intelligence layer is usually fixed and owned by the product. I don't want to pay twice for same LLM tokens.
I already use Claude Code and Codex, and I wanted my meeting assistant to use the same agent rather than requiring another AI subscription.
Speech-to-text/ASR works locally, and transcripts, notes, and project context are stored as local MD files. So, your coding agent will work with this files as usual, it's mean that under the hood, CallBro uses a full agent rather than a single summarization prompt and you can mange it by changing AGENTS.md or just command in chat.
It's still early, and I'm especially interested in feedback: What would you actually want a meeting agent to do after it has produced the notes?
Hi HN, I am building CallBro
The basic idea came from liking the bot free experience of tools like Granola, but not liking that the intelligence layer is usually fixed and owned by the product. I don't want to pay twice for same LLM tokens.
I already use Claude Code and Codex, and I wanted my meeting assistant to use the same agent rather than requiring another AI subscription.
Speech-to-text/ASR works locally, and transcripts, notes, and project context are stored as local MD files. So, your coding agent will work with this files as usual, it's mean that under the hood, CallBro uses a full agent rather than a single summarization prompt and you can mange it by changing AGENTS.md or just command in chat.
It's still early, and I'm especially interested in feedback: What would you actually want a meeting agent to do after it has produced the notes?
[dead]
[flagged]