Use cases

Use cases

Nine ways Deep Thought Cloud is used today: who it is for, which app or integration does the work, what you get and which plan it needs.

Each one is live today. Names in the examples are made up.

Customer meetings, recorded by rule

Who: A consultant or account manager with many customer meetings in Teams, Google Meet and Zoom.
Situation: Internal meetings should not be recorded. Every meeting with a customer should, without remembering to start anything.
How: Connect your Google or Microsoft calendar in the Web App. The account owner or an admin chooses the rule Meetings with external participants; before it is saved, the Web App shows how many meetings it would record in the next 7 days. Under Meetings, Upcoming, a single meeting or a series can be switched on or off. Step by step.
What you get: A participant named after you, for example "Kim notetaker", joins each of those meetings and records the audio. The meeting becomes a recording with transcript and speakers, and with Premium a summary, people, decisions and actions. You get an e-mail when the transcript is ready.
Plan: Premium and Enterprise. The meeting notetaker is in beta.
Try it: Become a tester to get the notetaker during the beta, then connect your calendar, step by step.

A meeting you were just invited to

Who: Anyone with the meeting notetaker, with or without a connected calendar.
Situation: A meeting link arrives in a chat five minutes before the meeting, or someone reads out a Meet code or a Zoom meeting ID.
How: In the Web App, under Meetings, Send to a meeting: paste the link, or type the Meet code or the Zoom ID (with its passcode). Teams needs the full link. If the meeting has a lobby, someone lets the notetaker in.
What you get: The same recording as above. The notetaker joins within seconds and can be removed from the same page.
Plan: Premium and Enterprise.
Try it: Become a tester, then send the notetaker to a meeting. See also customer meetings by rule.

Phone calls

Who: A team that sells or supports customers by phone.
Situation: Calls are recorded in the telephone system and should end up beside the meetings, with every line on the right person.
How: A call-recording integration, set up per account, brings the calls in. A two-channel call is transcribed one party per channel, and a known phone number names the other party. Which integrations are live.
What you get: Each call as a recording with the right speaker on every line, and the same summary, people and actions as a meeting.
Plan: Enterprise.
Try it: Ask us to set it up for your account. The live integrations.

Conversations and memos on the phone

Who: Someone who records on the move: a meeting in a room, an interview, a thought after a customer visit.
Situation: No meeting link and no computer, only the phone.
How: record-me on the iPhone (in beta, not yet publicly released; Android in testing) records and uploads, and the transcript comes back. The Web App records in the browser too, on a computer or a phone.
What you get: The transcript, the speakers and a title. With Premium, also the summary, the people and topics, and actions.
Plan: Free. Free keeps a recording's audio for 30 days; the transcript and the speakers stay.
Try it: Start free in the Web App and record in the browser (step by step). About record-me.

Voice memos into your notes on the Mac

Who: A Mac user who keeps notes in a folder or an Obsidian vault and records with Apple Voice Memos.
Situation: Memos pile up in Voice Memos and never reach the notes.
How: Trillian (alpha) watches Voice Memos on the Mac and uploads each new memo. With iCloud sync for Voice Memos on, memos recorded on an iPhone or an Apple Watch reach the Mac and are picked up too. Trillian writes each transcript into your folder.
What you get: Every memo as a note in your own files, and as a source in Deep Thought Cloud.
Plan: Free.
Try it: Download Trillian for macOS; it works on Free. Afterwards, ask across everything.

Ask across everything that was said

Who: Anyone with a few weeks of meetings, calls and memos in the account.
Situation: "What did Kim Berg say about the delivery plan for Exempelbolaget, and did we promise a date?" The answer is spread over several meetings.
How: Ask the chat under Insights in the Web App, or connect Claude Desktop, claude.ai or Claude Code over MCP and ask there: the ask tool answers in plain language with citations, and the other tools let the assistant search, follow a person and read a transcript. Connect an assistant.
What you get: One answer that points to the meetings and the lines it came from, instead of reading five transcripts.
Plan: Premium.
Try it: Become a tester for full access, then connect an assistant. It works best with a few weeks of meetings in the account.

Your AI assistant remembers what it has processed

Who: Someone who lets an AI assistant work through their recordings over MCP, for example turning each meeting into a note in their own files.
Situation: The assistant runs again tomorrow. It should take only what is new, not redo yesterday's meetings or miss one it never saw.
How: Every source can carry one note, the same for everyone in the account. When the assistant is done with a source it writes the note with the MCP tool set_source_note, for example "processed 2026-10-10, filed as Meetings/Exempelbolaget kickoff". Next time it asks list_sources for the sources without a note, and when it reads a source the note comes with it. Writing needs a connection approved with the kb:write scope; reading needs only the default scopes. Connect an assistant.
What you get: A record on each source of what an assistant has done with it, so the next run starts where the last one stopped. Writing the note never reprocesses the source or changes its transcript.
Plan: Premium.
Try it: Become a tester for full access, then connect an assistant and allow it to write. See also ask across everything.

Actions and follow-ups

Who: A project lead who runs several meetings a week.
Situation: What was agreed, and by whom, should not depend on someone's notes.
How: Actions, decisions and commitments are lifted out of every source and listed in the Web App under Actions, each with the source it came from. An AI assistant can add one over MCP. Actions can also go on to Todoist or ClickUp as tasks.
What you get: One list of open actions across meetings, calls and memos.
Plan: Premium; Todoist and ClickUp in Enterprise.
Try it: Become a tester for full access. Actions come from meetings and calls alike.

Your own system, through the API

Who: A developer who wants recordings in, or knowledge out, from their own system: a CRM, a case system, an internal tool.
Situation: Recordings already exist elsewhere, or the summary and actions should appear in your own system.
How: The REST API with a Bearer token: upload, processing status, sources with transcript and extraction, search and actions. API reference.
What you get: The same sources and extraction the apps use, in your own system.
Plan: Enterprise.
Try it: Read the API reference, then ask us about API access. Developers.

Private notes and a journal

Who: Someone who also records private thoughts: a journal, a reflection after a hard day.
Situation: Private recordings must not mix with work material.
How: Record with record-me or Trillian as above. A recording classified as personal never gets a work tag automatically, only as a suggestion, and a tag marked private stays out of shared summaries and outgoing e-mail. For the strictest case, local-only processing is available on request; the meeting notetaker is then not available.
What you get: A searchable record of your own thinking, kept apart from work.
Plan: Free for the recording; local-only processing on request.
Try it: Start free in the Web App, or ask about local-only processing.

Start free in the Web App, become a tester to try the meeting notetaker in beta, set up the meeting notetaker, see the integrations, or compare the plans on Pricing. Stuck? Help.