Privacy policy

Atlas Meeting (“Atlas,” “we,” “us”) provides an AI assistant for your meetings — on your Mac or joining your video calls — that takes notes, renders visuals, and produces summaries. This policy explains what data we collect, how we use it, and the choices you have. It applies to atlasmeeting.com, the Atlas desktop app, and the Atlas service.

The short version: we never store your meeting audio, we collect the minimum needed to run the service, we do not sell your data, and your meeting content is never used to train AI models — by us or by our providers.

Account information. When you sign up we collect your name, email address, and authentication identifiers through our auth provider (Clerk). If you sign in with Google, we receive your basic Google profile and email address.

Google user data. With your explicit consent, Atlas accesses your Google Calendar events (read-only) to know when meetings are scheduled and who is attending. We request only the scopes needed for these features and never request write access to your calendar.

Meeting data. Atlas stores the transcript, summary, decisions, action items, and the visual artifacts it generated. Your audio is captured solely to produce the transcript: it is transcribed in real time and is never stored — not by Atlas, and not by our recording processor, which we operate in zero-data-retention mode. At no point does a recording of your meeting exist on our servers.

Connected tools. If you choose to connect workplace tools (Gmail, Google Drive, Google Calendar, GitHub, Linear, Slack, Notion), Atlas requests the minimum read-only permissions needed to bring that context into your meetings. The OAuth tokens for these connections are held in dedicated credential vaults — never in our database — and you can disconnect any tool at any time from Settings.

Operational data. Standard logs (timestamps, error traces, aggregate usage) used to operate and secure the service. These do not contain meeting content. Product analytics are consent-gated, limited to feature-usage metadata, and never include meeting content.

We use the Google Calendar data you authorize solely to provide user-facing features of Atlas: identifying upcoming meetings so the assistant can join, resolving meeting attendees to send recaps, and showing your schedule inside the dashboard.

We do not use Google user data for advertising, we do not sell it, and we do not transfer it to third parties except as needed to provide the service (see Subprocessors) or as required by law.

Atlas’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

Specifically, we do not use Google user data for serving advertisements, we limit human access to this data except as required for security, to comply with applicable law, or where you have provided explicit consent, and we do not use or transfer this data for purposes unrelated to providing or improving the Atlas features you requested.

All data is encrypted in transit (TLS 1.3) and at rest (AES-256, under keys managed in AWS KMS with automatic rotation). Atlas runs on AWS servers located in the United States.

Your meetings are private by default: only you can see them unless you explicitly share one. Our database stores no usable credentials — OAuth tokens are held in dedicated vaults (our authentication provider, AWS Secrets Manager, or the connector provider), and the desktop app keeps no meeting content or credentials on your device’s disk.

Atlas employees do not read your transcripts or summaries. Engineering may access aggregate system logs to diagnose issues, never the content of your meetings.

Atlas uses AI to turn your meetings into notes, summaries, and visuals. Meeting content is processed by Anthropic’s Claude models running inside Atlas’s own private AWS environment (Amazon Bedrock) — it is not sent to a shared consumer AI service.

We do not train AI models on your data, and we do not allow anyone else to either: AWS and Anthropic do not use data processed through Amazon Bedrock for model training. Your meetings improve your meetings — not anyone’s models.

We share data only with the service providers needed to operate Atlas: AWS (compute, storage, and AI inference via Amazon Bedrock; US region), Anthropic (LLM inference via Bedrock; never used for training), Deepgram (real-time transcription), Recall.ai (meeting capture, operated in zero-data-retention mode — it stores nothing), Clerk (authentication), Composio (connector OAuth vault), Resend (delivery of the recap emails you send), PostHog (consent-gated product analytics; never meeting content), and Vercel (web hosting). We never sell your data or share it for advertising.

A current list with locations and data categories is available on request at trust@atlasmeeting.com.

Deleting your account permanently deletes all of your data — meetings, transcripts, notes, artifacts, uploaded files, and connections — immediately, with encrypted backups aging out within 7 days. There is never any audio to delete, because none is stored. To delete specific meetings or make any other data request, email trust@atlasmeeting.com and we will act on it promptly.

You can revoke Atlas’s access to your Google data at any time from your Google Account permissions page, or by disconnecting Google in Atlas Settings.

We may update this policy as the service evolves; material changes will be reflected by the “last updated” date below and, where appropriate, an in-product notice.

Questions, data requests, or DPA paperwork — email trust@atlasmeeting.com.