Mic & system audio, captured cleanly.
Your mic, plus whatever's playing through your Mac — Zoom, Meet, Teams, a podcast. No virtual driver hacks.
Boswell records your meetings, transcribes them on your Mac, and writes everything to a Markdown folder you own — speakers and action items. Your audio doesn't leave the machine unless you say so.
Boswell transcribes your meetings — no bots in your calls.
Your mic, plus whatever's playing through your Mac — Zoom, Meet, Teams, a podcast. No virtual driver hacks.
Tiny through Large v3, your pick. Models live on disk. The audio stays on your Mac while it becomes text.
Speaker labels run on-device. You can rename "Speaker 2" to a real name once and it sticks.
Apple Intelligence handles the summary on-device — no API key needed. Bring your own (Anthropic, OpenAI, Ollama) if you want a different LLM.
Search every transcript. Find a quote, a topic, a decision — or ask for a summary across them all.
Everything writes to a folder you choose — plain .md with frontmatter, links, and a sidecar audio file. Drop the folder in Obsidian and it just works.
Boswell records, transcribes, labels speakers, and writes the summary — all while the meeting is happening. Six minutes of automatic work, instead of twenty minutes writing it up after.
Boswell records the mic and system audio, transcribes with Whisper, and labels who's speaking. You don't have to switch apps or watch a progress bar.
A Markdown file lands in the folder you chose. Title, attendees, summary, action items, full transcript, and the audio. Open it in any editor.
Boswell ships with all the cloud features off. The mic, the model, the speaker labeling, the summary — all on-device. Want a cloud LLM to write the summary? Plug in your key. We don't store it, route through it, or see what comes out.
Universal binary, around 38 MB. First launch, it asks for mic and screen-recording permission — that's it. No account, no login.
From anywhere — meeting, hallway, walking. The on-air dot in your menu bar shows it's recording; press again to stop.
"What did we decide about the migration?" — Boswell finds the quote and the file. Open it in your editor of choice.
We're not going to name names. You know the ones — the apps that join your call as a fifth attendee, screenshot your slides, and send you a "happy summary".
Try the free tier on this week's meetings. When you're ready for system-audio capture, who said what, and the dictation hotkey, $49 unlocks it. Forever — same machine, same person, no expiry.
A real free tier — enough to see whether Boswell fits the way you actually work.
The whole app — meeting capture, who said what, dictation, transcript search, the works.
No. Boswell captures audio directly from your Mac — your microphone and whatever is playing through your speakers. Other attendees see nothing new. There is no bot, no email invitation, no "is it okay if I record this" awkwardness.
By default, no. Whisper runs locally; speaker labels run locally; Apple Intelligence handles summaries on-device; embeddings are stored in a SQLite file on disk. If you bring your own cloud LLM key — Anthropic, OpenAI, Ollama — that traffic goes directly from your machine to the provider, never through us.
Your transcripts are plain Markdown in a folder you chose. They'll keep opening in any text editor — no Boswell required.
Yes, with caveats. Boswell needs Microphone and Screen Recording permissions; some MDM profiles block one or both.