Export Your Data

You can export a copy of your Groupthink data at any time, on any plan, including the free plan. Groupthink prepares an archive and emails you a download link.

There are two exports, both in the same place:

Export	Button	What you get
Export your data	Export my data	Everything you put in, and the meetings you host with their transcripts. Summaries and other things Groupthink wrote from them are not in the file.
Just your captures	Export my captures	Only the things you said or typed to Groupthink directly, as Markdown by day plus JSONL.

Both are free. Neither changes or deletes anything in your account.

Where to find them

1. Click your profile avatar and choose User settings, then open the Account tab. Or go straight to app.groupthink.com/settings#account.
2. Find the Export your data card. The Export my data button is at the top. Below the divider, under Just your captures, is the Export my captures button.
3. Click the one you want. The card shows a message that your export is being assembled. You can leave the page; it runs in the background.
4. When it is ready, you get an email with a Download your data button (Download your captures for the captures export). Click it to download the archive. The email subject is "Your Groupthink data export is ready" or "Your Groupthink capture history is ready".

You can run both exports at the same time. Each one shows its own status on the card.

How long the download link lasts

The link in the email works for 48 hours and can be used once. After you download the file, or after 48 hours, the link stops working and opens a page that points you back to settings. Start another export if you need a fresh link.

Do not forward the email. The link opens your data.

What is in the full export

The Export my data archive contains two files:

• export.json: a structured copy of your account data, in the sections listed below.
• relationships.csv: one row per person, in a spreadsheet format you can open in Excel, Numbers, or Google Sheets. The columns are name, primary_email, phone, relationship_role, relationship_tier, desired_frequency, birthday, created_at, and current_roles.

Sections of export.json

The file has eight sections, in this order:

Section	What it holds
meta	Facts about the file itself: schema_version, generated_at, account_id, and two plain-English notes, notice and meetings_scope, that describe what the archive covers.
account	Your own profile: name, email, bio, job title, company, location, website, social links, and when the account was created.
relationships	One entry per person, with their details and lists of emails, github_logins, facts, open_threads, gift_ideas, intentions, interventions, artifacts, and dossier_entries.
projects	Your projects, with their stages.
goals	Your goals.
ingested_notes	Your notes, each with its title, content, and source. Each note also has channel, source_app, and captured_via, the same three fields described under the captures export.
event_personal_notes	Your personal notes on calendar events and meetings.
meetings	Meetings on agendas where you are a host, each with its transcript.

Meetings in export.json

Each entry in meetings is one meeting on an agenda where you are a host. Meetings you attended on someone else's agenda are not included. It has these fields:

Field	What it holds
id	The meeting ID.
agenda_id, agenda_name	The agenda the meeting belongs to.
name	The meeting headline.
source	Where the meeting came from.
starts_at, ends_at	The scheduled start and end.
started_at, ended_at	When the meeting actually started and ended.
created_at	When the meeting record was created.
attendees	The people in the meeting who have Groupthink accounts, each with id, name, and email.
recap_recipients	People with Groupthink accounts who were sent the recap and were not attendees, each with id, name, and email.
pending_recap_recipients	Addresses the recap was sent to that have no Groupthink account yet. Each has email only.
recap_sends	One row for each time the recap went out on each channel, oldest first, with recipient_type, email, name, user_id, channel, sent_at, and confirmed_at.
unknown_participants	Participants the video call reported that were not matched to a Groupthink account, as the call provider sent them.
transcript	The transcript, one entry per sentence, in meeting order. The list is empty when the meeting has no transcript.

Each sentence in transcript has id, speaker, speaker_user_id, speaker_relationship_id, text, relative_started_at, relative_ended_at, started_at, ended_at, and created_at. text is the words as transcribed. speaker_user_id matches an id in that meeting's attendees or recap_recipients.

What is in the captures export

The Export my captures archive is built to be read by a person and parsed by a tool:

• README.md: explains the layout and the source labels.
• captures/YYYY/YYYY-MM-DD.md: one Markdown file per day, in your own timezone. Each capture is a heading with the time, the source, and who it was filed to, followed by the exact text you captured. When the desktop app recorded which app the text went into, the heading names it, for example voice in Slack.
• captures.jsonl: the same captures, one JSON object per line, for tools. Each line has id, captured_at, source, source_raw, channel, source_app, captured_via, to_entity, about_entities, text, and role.

Three of those fields say where a capture came from. Each is null when it does not apply:

Field	What it holds
channel	How you said a logged interaction happened, such as call, text, email, or in_person.
source_app	The app the desktop app pasted your dictation into, such as Slack. For a capture that came in through an MCP connector, it is the name the connector gave for itself.
captured_via	Which client a quick capture arrived through: desktop, web, or mcp.

A day file looks like this:

# 2026-08-20 — Thursday, August 20, 2026

_Times are America/Los_Angeles._

### 14:05 — voice → Theo Marchetti

Theo is worried about Q3 pipeline. Follow up Friday.

### 16:30 — ask → Theo Marchetti

**You asked:**

What does Theo care about right now?

**Groupthink answered:**

Theo cares about landing Q3.

Each block ends with a hidden HTML comment carrying the capture ID. You will not see it when the file is rendered. It is how GitHub repo sync knows not to write the same capture twice.

What counts as a capture

A capture is something you said or typed to Groupthink yourself:

Source label	What it is
voice	A voice note you recorded, transcribed
router	A quick capture typed into the capture box and routed to a person
quick_log	A Quick Log line about a person
note	A note you typed or pasted on a person's page, in the capture inbox, or from the iPhone app
text	A text conversation you pasted or forwarded in
ask	A question you asked Groupthink about someone, with the answer Groupthink gave

A question you asked the AI includes both sides, labelled. A voice note whose transcript never arrived is still listed, with a placeholder instead of text.

What is not in the captures export

Content that came in from sources you connected or forwarded: email, Slack, Google Drive, screen content, GitHub activity, and uploaded meeting decks. That content is in the full Export my data archive. What the Advisor wrote to you is not in either export.

Notes

• Exporting does not change or delete anything in your account. It is a read-only copy.
• If the card says your last export did not finish, click the button again. It now reads Export again (Export captures again for the captures export). If it keeps failing, email support.
• Want your captures and meetings kept up to date in a repo you own, without exporting by hand? See GitHub repo sync (Pro).
• If you instead want to close your account, see Close Your Groupthink Account. After 30 days a closed account, with its contacts and notes, is permanently deleted, so export first if you want a copy.

Related

• GitHub repo sync
• Set up repo sync
• Close Your Groupthink Account
• Managing Your Billing & Subscription