Look at David Shapiro's project on GitHub, not Raven but the other one that is more fleshed out. He already does the summarization of dialogue and retrieval of

Look at David Shapiro's project on GitHub, not Raven but the other one that is more fleshed out. He already does the summarization of dialogue and retrieval of relevant info using the OpenAI APIs I believe. You could combine that with the Chrome web speech or speech-to-text API which can stay on continuously. You would need to modify it a bit to know about third party conversations and your phone would run out of battery. But you could technically make the code changes in a day or two I think.