Screenshots every 5 second for richer context and summaries
I think it would be a great idea if other than audio, the talat system can take screenshots every 5 seconds (or configurable) so that when the meeting is getting summarized/processed, the local or my subscription LLM can read the screeshots as well for more accurate meeting notes etc. Would be...
Codex agent backend: let me use my own Codex config (custom model and provider)
I use Talat with an OpenAI-compatible gateway on localhost. Summaries work well through a custom LLM profile, but the Codex agent backend can't reach the same gateway. From what I can see, Talat runs Codex with --ignore-user-config and only allows models from a built-in list (GPT-6...
Right now Talat only imports audio and video files. I'd like to import existing transcripts and notes as text or Markdown too, for example exports from Granola, Otter or Fireflies. Many people moving to Talat have months of meeting history in another tool, usually as text with no audio. Importing...
Right now Talat merges speaker and mic into the same audio channel. This unnecessarily removes data that can be useful for future reprocessing and speaker assignment. Please allow the recording to save the speaker and mic onto different channels in the recording.
I would like to add photos from disk or from clipboard from the people that i are speaking. Also i would like to attribut a company to them also with a logo so that i have more info and see who is speaking where and from what company which arguments came in a discussion. If the email is known...
I would like to split a meeting already during the recording (cause something went wrong and two calls are mixing). Also i would like to change the meeting to another one (having parallel meeting invitations and it selected the wrong one)
When you copy all actions and their owners, it automatically adds some non markdown tickbox. Usually I’m pasting these action points into a system which deals with markdown properly such as Slack, Notion or Docs. - [ ] Talk to x, y and z (Owner Name) I would love this to be copied as bullet...
The transcription quality for german ist pretty bad. So bad, that I have to send basically every meeting to another transcription model. I tested quiet a few models for meetings in German with several speakers (and also a lot of tech / english words), the following models performed really well...
When I activate the dictation screen I would like to be able to select which prompts should apply from my currently configured prompts in Dictation Settings
I can currently filter meetings by tags with is very useful, it would also be very helpful, as part of my routine tidy up, to be able to show meetings that haven’t been tagged. As part of this tidy up I would also like to be able to show meetings where the summary hasn’t been created.
I know it’s likely a thing for a few of us only, but often I am meeting with people or teams across companies. It would be awesome to select a company and only the people under that company are auto tagged if that makes sense? Quite often it believes someone is speaking (or continues to prompt)...
Sometimes i have meeting after meeting. And suddenly its in the first started meeting and i forgot before to switch the meeting. So it would be good to link and separate a meeting to another calendar entry or create a new meeting while its still running. so that the participant list (when...
I have a regular meeting with the same people, and adding each participant by hand every time is tedious. Groups already exist for vocabulary. Please let me add a whole group to a meeting's participants in one action from the who's here panel. Bonus: remember the group for a recurring calendar...