Post Transcribe Pipelines

Beta installed! Will report back.

Hey Kyle, just following up - did you face any issues or did everything feel stable? Trying to push for general release before christmas and haven't heard any bug reports anymore so far.

Hey Alex — sorry for the delay. I did download and test the beta, but I stepped away for the holidays. Overall, the beta worked well for me, and I really liked the increased refinement of the UI and pages. I do have a couple of notes: - It wasn’t very clear which transcribe profile was active at any given time, aside from the master toggle in Settings. It would be great to be able to see and toggle profiles directly from the menu bar panel, including selecting which profile is active.
 - It would also be useful to be able to ad‑hoc run previously recorded clips through different profiles. Outside of that, one thing I was personally hoping for was the ability to use cloud models via BYOK support. I understand that a core value of the app is its offline and local-first approach, but for my workflow, I’d much prefer doing STT locally and then optionally transcribing via higher‑quality cloud model providers. I’ve spent a lot of time working with offline models in other workflows and coding projects, and I think they’re generally underrated. However, for this particular use case, they can slow things down. For users who aren’t as familiar with the tradeoffs, this might lead them to feel that the app itself is slow or bloated. Even with the smallest model, the app uses around 2–3 GB of RAM during transcription and holds onto that memory for a while. On my MBP with 18 GB of RAM—running tools like Cursor or Warp alongside a Chromium‑based browser—that’s a tough tradeoff for always‑ready local transcription. This may just be my personal preference, but I think the best compromise would be supporting both local and cloud‑based options.