Privacy Policy

Last updated November 13, 2026

This Privacy Policy explains how WeMachines (“WeMachines”, “we”, “us”, or “our”) collects, uses, shares, and protects information when you use our website at wemachines.com, our application at app.wemachines.com, and related features (together, the “Service”). It works alongside our Terms of Service.

In short: your team’s content is yours. We process it to run the workspace and the features you choose to use, we rely on a small set of trusted providers to do that, and we don’t sell your data.

1. Information we collect

Access requests

Until November 3, 2026, WeMachines was invite only. If you asked for access on our website then, we keep the work email and name you gave us, any answers you added about the tools your team uses, the site that sent you and the country your request came from, which we used to decide on your request and to email you when you were let in.

Account information

When you sign up, we collect your email address (used to identify your account) and basic profile details you provide, such as a display name, avatar and banner image. We do not store passwords: sign-in is passwordless, using either a one-time code sent to your email or your Google, Apple or Microsoft account.

If you choose “Continue with Google”, Google confirms your identity to us and shares your email address and basic profile information (such as your name and profile picture). We never receive your Google password, and signing in asks for nothing beyond that basic profile. Google Calendar and Gmail are the only other Google services we ever request. Each is entirely optional, each is its own separate permission, and each is requested only if you connect it yourself in Integrations (see “Google Calendar and Gmail integrations” below). We never request your Drive or your files. Google’s own handling of that sign-in is governed by its privacy policy; you can review and revoke WeMachines’s access at any time in your Google account settings.

If you choose “Continue with Microsoft”, which takes work and school Microsoft accounts only, Microsoft confirms your identity to us and shares your email address and basic profile information (such as your name). We never receive your Microsoft password, and signing in asks for nothing beyond that basic profile. Your Outlook mailbox is a separate permission, requested only if you connect Outlook yourself in Integrations. Microsoft’s own handling of that sign-in is governed by its privacy statement and your organization’s agreement with Microsoft.

Workspace content

We store the content you and your team create in a workspace, including chat messages, tasks, documents, polls, comments, reactions, and uploaded file attachments. This content is provided by you and shared with the members of your workspace.

Integration data

If you connect integrations, we process the data needed to make them work. For GitHub, this includes repository content and metadata accessed under the permissions you grant. For Slack, this includes the messages in the channels the bot is in, who is who in your Slack, and the Slack profile email used to match a Slack account to a member of your workspace - see “Slack and Discord integrations” below. For Discord, this includes the messages in the channels machines learn from and which Discord account each member linked. For Google Calendar, this includes the events on the calendar you connect, and for Gmail, the email you ask your personal AI teammate about and the email conversations you write in, which it learns from - see “Google Calendar and Gmail integrations” below.

Billing information

When you subscribe to a paid plan, payment is processed by our payment provider, Stripe. Stripe collects and processes your payment details (such as card or other payment information) and billing address directly; we do not receive or store full payment card numbers. We receive limited billing records from Stripe, such as your plan, subscription status, the last digits and type of your payment method, and invoices, so we can manage your subscription.

Voice, video & transcripts

When you use huddles or the voice machine, we and our real-time communication providers process audio and video streams, and may generate captions, transcripts, and synthesized AI voice as part of the feature.

Usage, device & log data

Like most online services, we and our providers automatically collect technical information such as IP address, browser and device type, pages and features used, timestamps, and diagnostic logs. We use this to operate, secure, debug, and improve the Service. For the same purposes we use product analytics, which record which pages and features are used.

We do not use session replay. Our analytics record which pages and features are used, not a reconstruction of your screen, and not what is on it: in the app, the text of what you click and the page's title (which names the task or page open) are removed before anything is sent. We previously did use session replay, and stopped on August 14, 2026: a replay reconstructs the page as it appeared to you, so it could include your workspace's content and the names of people in it, and the person who agreed to analytics in their browser is not the person that content belongs to. All recordings made before that date have been deleted at our analytics provider.

Cookies & local storage

We use browser storage (local storage and IndexedDB) to keep you signed in, remember preferences, and support offline and real-time editing of documents. Your sign-in session is held in local storage, not in a cookie. We do not use advertising cookies, and apart from the visitor identification on wemachines.com described below, which runs only if you accept it, we do not track you across other websites.

Analytics need your permission. PostHog stores an identifier in a cookie so repeat visits count as one person rather than many. On wemachines.com the same permission also loads Apollo's website tracker, which tells us which companies visit: it keeps an anonymous identifier in your browser and sends Apollo the pages you view, your IP address and the campaign that brought you. Apollo may also load a resolver from its partner LiveIntent, which can match your browser to a hashed form of your email address that LiveIntent already holds, and which stores identifiers of its own. The same permission loads lemlist's website tracker too, which tells us the same thing without storing anything in your browser: it sends lemlist the pages you view, how long you stay on each and how often you click, with your IP address and browser type, and lemlist works out the company from the IP address. We ask before any of this, and nothing is set or loaded until you accept - decline and analytics never start. A second cookie records that answer so you are not asked again; it holds nothing but your choice and is set either way. That cookie and PostHog's are set on wemachines.com, so one answer covers the marketing site and the app, and your choice is remembered for six months.

The same permission lets wemachines.com keep a note in local storage of how your first visit reached us: a campaign tag or the site that linked here, the page you arrived on, and when. It reaches us only inside a form you choose to send, such as Book a demo. Without your permission the note lasts only while the site is open, and nothing is stored.

The remaining cookies are strictly necessary, so they are not part of that choice and cannot be used to identify you elsewhere: a short-lived cookie that stops the GitHub and Spotify connection flows from being hijacked, and, where your workspace runs an app, a cookie on that app's own address that tells the app who you are. Neither can be read by page scripts.

You can change your mind at any time: use at the bottom of any page here, or Settings → Profile → Analytics in the app. Turning analytics off also clears what has already been stored in your browser on wemachines.com; cookies LiveIntent keeps on its own domain are cleared from your browser's settings. Blocking cookies in your browser works too, though the sign-in and editing features above rely on browser storage and will not work without it.

The sign-in screen and the book-a-call form load Cloudflare Turnstile to tell people from bots. It is a third party that may set its own cookies, and it loads only on those two pages.

The sign-in screen also loads Fingerprint, which recognises the browser and device a new account is created from, so that one person or one script cannot create many accounts. It stores an identifier in your browser, and it loads only on the sign-in screen.

2. How we use information

  • to provide, maintain, and operate the Service and its features;
  • to authenticate you and keep your account and workspace secure;
  • to power AI features you use, including generating responses, summaries, and code changes;
  • to enable integrations you connect, such as GitHub, Slack, and Google Calendar;
  • to respond to support requests and communicate with you about the Service;
  • to monitor, debug, prevent abuse of, and improve the Service; and
  • to comply with legal obligations and enforce our Terms.

Where required by law, we rely on the following legal bases: performance of our contract with you (to provide the Service), our legitimate interests (to secure and improve the Service), your consent (where requested), and compliance with legal obligations.

3. AI processing

When you use AI features, relevant workspace content is sent to AI providers to generate a response, and, for voice features, to speech-to-text and text-to-speech providers. AI requests go through OpenRouter, our AI gateway, which passes each one to an endpoint that runs the model. Where a model has endpoints that keep no data (zero data retention), the request is pinned to them and fails rather than going anywhere else, and every model WeMachines' built-in AI uses has them. WeMachines selects the model for every AI feature, and changes it as better ones come out; nobody in a workspace picks one. Today AI teammates run on a model built on DeepSeek's V4.1 Flash or on Moonshot AI's Kimi, each of which OpenRouter runs on an independent inference provider that keeps no data, or on Anthropic's Claude on Amazon Bedrock and Google Vertex AI, depending on the level a team sets for each AI teammate, with OpenAI's GPT on Microsoft Azure standing in when Kimi is unavailable; summaries and most other everyday AI features run on DeepSeek's V4.1 Flash model, which only Cloudflare (Workers AI) runs for us, keeping no data and none of it going to DeepSeek's own service, and which also stands in for the first of those models when it is unavailable; quick, high-volume tasks run on Google's Gemini on Google Vertex AI, which also stands in when a model is unavailable; some in-depth background work runs on OpenAI's GPT on Microsoft Azure; and coding runs, app builds and pull request descriptions run on Anthropic's Claude on Amazon Bedrock and Google Vertex AI. What is said on a voice call with an AI teammate goes to DeepSeek's V4.1 Flash model, run by independent inference providers that keep no data. Repository studies - the Study button on a team's connected repositories, which an admin starts - run on Anthropic's Claude on those same endpoints, so the source files a study reads are sent to them. When an AI teammate researches the web, its questions are sent to LinkUp, our search provider, which keeps no data (zero data retention). When it looks up growth data - public social profiles and posts, search results, map listings, ad libraries, reviews, launches, job posts and site traffic - what it looks up (account names, search words or links on those platforms) is sent to Apify, which collects the public data. AI teammates also remember what their team says: when a conversation in a room, or with an AI teammate, has gone quiet, the default model above reads it and writes down what is worth remembering as short notes, kept in the workspace with the words of the messages they came from, and AI teammates recall them when they answer. Each note is turned into an embedding, a list of numbers that lets notes be found by meaning, by OpenAI's embedding model on Microsoft Azure, which keeps no data. Whispers between two people are never read for this. A conversation with your own personal AI teammate is read the same way, but what it writes down goes on your private memory shelf: only you can read it, only your personal AI teammate recalls it when it answers you, and it is deleted with your membership. When memory first reads a workspace it goes back 30 days. We do not use your content for AI training, and we rely on our AI providers' commitments not to train their models on data submitted through their business APIs.

4. Code & repository data

If you connect a GitHub repository, our build and coding-machine features may clone your repository into a secure cloud environment to run, edit, preview, and propose changes, and may send relevant code to AI providers to generate those changes. Proposed changes are delivered as pull requests for your review. Your source code remains in your own GitHub repository, and you can disconnect the integration at any time.

5. Slack and Discord integrations

If a workspace admin connects Slack, we store your Slack team’s id, name and address, the bot’s user id, the granted scopes, and an access token for the bot - encrypted at rest, and readable only by our servers.

What your machines learn from Slack. When an admin first connects Slack, learning is switched on and the bot joins your Slack’s biggest public channels (at most five, each with two or more people). It joins any other public channel only when an admin picks it (one by one, or all of them), and an admin can switch learning off for a channel or for all of Slack. We store the messages people write in the channels it learns from, threads included (not bots’ messages, and never private channels, direct messages or group DMs), for 90 days, and a list of your Slack’s public channels and of the people in your Slack (name, title, timezone, and which member of your workspace each one is, matched by email - the email itself is not stored). When a channel is picked, the bot reads up to its last 30 days. Your workspace’s AI teammates use this to answer questions about what was said in Slack and to learn how your team works: what they learn is stored as part of your workspace’s machine memory, which every member can read and any member who can edit can delete. This content is processed by our AI providers to do that and is never used to train a model.

When someone mentions the bot, we receive that mention and the surrounding messages (the thread, or the last few messages in the channel), together with the Slack profile email of the person asking, which is used to match them to a member of the connected workspace. That content is processed to answer or to file a task, including by sending it to our AI providers. How much of your WeMachines workspace the bot may say back out in Slack is an admin setting that starts at nothing.

An admin can switch learning off for one channel or for all of Slack in Integrations → Slack; doing so, removing the bot from a channel, disconnecting Slack or uninstalling the app deletes what we stored from it and what your AI teammates learned from it. Disconnecting also revokes the token and deletes the connection. Slack’s own handling of the messages and profile data in your Slack workspace is governed by Slack’s privacy policy and your agreement with them.

Discord

A workspace admin can connect one or more Discord servers - the team’s own, or a community it runs. We store each server’s id, name and icon, and the admin’s settings for it. Each member who wants the bot to answer them links their own Discord account from WeMachines; we store that account’s id and name, and Discord does not share its email with us. One bot serves every connected server, and its token is never stored per server.

What your machines learn from Discord. Unless an admin switches it off, we store the messages people write in the channels machines learn from - every channel the bot can see in a team’s own server, and only the channels an admin picks in a community server - threads included (not bots’ or webhooks’ messages, and never direct messages), for 90 days, with who wrote them. When a server or channel is switched on, the bot reads up to the last 30 days of it. Your workspace’s AI teammates use this to answer questions about what was said and to learn how your team and community work, stored as part of your workspace’s machine memory. This content is processed by our AI providers to do that and is never used to train a model. A message deleted in Discord is deleted here too, with what was learned from it.

When someone asks the bot (a mention, /ask, or a message’s Apps menu), we receive the message and the few before it, and answer only if the person linked their Discord account to a member of the connected workspace; everyone else is ignored. An admin decides per server whether the bot answers at all - a community server starts with it off - and how much of the workspace it may say there.

An admin can switch learning off for one channel or a whole server in Integrations → Discord; doing so, removing the bot from the server or disconnecting the server deletes what we stored from it and what your AI teammates learned from it. A member can unlink their own Discord account at any time. Discord’s own handling of the messages in your server is governed by Discord’s privacy policy and your agreement with them.

Telegram

A workspace admin can connect Telegram groups the team runs - most often its community - by adding our bot, @WeMachinesBot, to the group. Only an admin of that group can add it, and the bot posts nothing there. We store each group’s id, title, public username if it has one, member count and topic names, and the admin’s settings for it. One bot serves every connected group, and its token is never stored per group.

What your machines learn from Telegram. Unless an admin switches it off, we store the messages people write in a connected group from the moment the bot joins - never anything said before, never bots’ messages, and never private chats with the bot - for 90 days, with each sender’s Telegram name and user id. Your workspace’s AI teammates use this to answer questions about what was said and to learn how your community works, stored as part of your workspace’s machine memory. What people in a community say is treated as information, never as an instruction to an AI teammate. This content is processed by our AI providers to do that and is never used to train a model. The bot does not answer in the group.

Deleting it. Anyone - whether or not they use WeMachines - can send /forget to @WeMachinesBot in Telegram, and we delete every message of theirs we store, in every workspace, with what AI teammates learned from them. An admin can switch learning off for a group or disconnect it in Integrations → Telegram, and removing the bot from the group does the same; each deletes what we stored from that group and what your AI teammates learned from it. Telegram does not tell bots when a message is deleted, so a deleted message stays until its 90 days pass or one of these deletes it. Telegram’s own handling of your messages is governed by Telegram’s privacy policy.

Microsoft Teams

A workspace can connect a Microsoft Teams team: someone in the company adds the WeMachines app to the team in Microsoft Teams, and a workspace admin connects it with a one-time code from Integrations → Microsoft Teams. The person who adds the app grants it permission to read the messages in that team’s channels, and the bot says in the team what it does there once it is connected. We store the team’s id, name and Microsoft tenant id, its channels’ names, and the admin’s settings for it. One app serves every connected team.

What your machines learn from Microsoft Teams. Unless an admin switches it off, we store the messages people write in the connected team’s standard channels - and in a private or shared channel only once someone in it adds the app there - from the moment it is connected - never anything said before, never bots’ messages, never group chats, and never chats with the bot - for 90 days, with each sender’s name in Teams and Microsoft user id. Your workspace’s AI teammates use this to answer questions about what was said and to learn how your team works, stored as part of your workspace’s machine memory. This content is processed by our AI providers to do that and is never used to train a model. The bot does not answer in the team.

Deleting it. Anyone in a connected team can open a chat with the WeMachines bot in Microsoft Teams and send forget, and we delete every message of theirs we store, in every workspace, with what AI teammates learned from them. A message deleted in Teams is deleted here too. An admin can switch learning off for a team or disconnect it in Integrations → Microsoft Teams, and removing the app from the team does the same; each deletes what we stored from that team and what your AI teammates learned from it. Microsoft’s own handling of your messages is governed by Microsoft’s privacy statement and your organization’s agreement with Microsoft.

6. Google Calendar and Gmail integrations

Connecting Google Calendar is optional and is never requested at sign-in. If you connect it in Integrations → Google Calendar, we ask Google for a single scope, https://www.googleapis.com/auth/calendar.events, which lets us view and manage events on your calendar and nothing else. Connecting the calendar does not give us access to your Gmail (a separate connection, described below), your Drive, your contacts, your files, or your calendar’s sharing and account settings. The resulting access and refresh tokens are stored encrypted at rest and are readable only by our servers.

We use that access only to carry out what you ask your AI teammates to do: read your upcoming events, work out when you are free, and create, update or cancel an event on your behalf. To do that, the relevant event details are sent to our AI providers to generate the response, exactly as described in “AI processing” above. We do not copy your calendar into our database, and we do not read it in the background. Your AI teammates can use it only in the teams you turn it on in, and an invitation to someone outside the team is sent only once you allow it. Disconnecting it from every team in Integrations → Google Calendar deletes the stored tokens; you can also revoke access at any time from your Google account settings.

Gmail

Connecting Gmail is optional, is never requested at sign-in, and is a separate permission from Google Calendar: you can connect either one without the other. If you connect it in Integrations → Gmail, we ask Google for two scopes: https://www.googleapis.com/auth/gmail.readonly, which lets us search and read your email, and https://www.googleapis.com/auth/gmail.compose, which lets us save a draft in your Gmail and send email from it. The resulting access and refresh tokens are stored encrypted at rest, separately from your calendar’s, and are readable only by our servers.

Only your personal AI teammate - the one in your workspace that only you can see, talk to and read - can use your connection, and it uses it in two ways. The AI teammates the workspace shares never can, and neither can any other member of your workspace.

First, when you ask your personal AI teammate about your email: to search for messages, read the conversation you point it to, write the draft you asked for, and send the email you asked it to send. An email goes out on its own only to members of your workspace, to you, or to an address you wrote in your request; an email to anyone else is saved as a draft and sent only when you press Send on it yourself, in WeMachines or in Gmail. The relevant messages are sent to our AI providers to generate the response, exactly as described in “AI processing” above. An answer drawn from your email is stored as a chat reply that only you can read, and is deleted with the conversation it belongs to. For each email sent we keep only how many people it went to, to limit how much can be sent.

Second, so that your personal AI teammate knows your work the way you do, it learns from the email conversations you take part in. Every few minutes it lists the messages you sent in the last 30 days, reads the conversations they belong to - your messages and the replies in them, a few conversations at a time - and writes down what is worth remembering, such as who you work with and what you agreed, as short notes on your private memory shelf (see “AI processing” above), kept with the words of the messages they came from. Only you can read them, and only your personal AI teammate recalls them, when it answers you. Before a conversation is read, codes, passwords, keys, card numbers and sign-in and reset links are removed from it, and a conversation about anyone’s health, sex or love life, beliefs or identity, legal trouble or money trouble is not learned from at all. Email you only received, in a conversation you did not write in, is never read in the background, and nothing is drafted or sent in the background.

Once your personal AI teammate has read your email while answering you, it cannot search the web, fetch a web page or use a browser for the rest of that answer, so nothing from your email reaches a search provider or a website. Your Gmail is never available to other AI apps you connect to WeMachines, for example over MCP: only your personal AI teammate, running on the AI providers described in “AI processing”, reads it. Your personal AI teammate can use your connection only in the teams you turn it on in. Turning Gmail off for a team in Integrations → Gmail stops both uses there and deletes what your personal AI teammate learned from your email in that team; disconnecting it from every team also deletes the stored tokens, without affecting your calendar connection. You can also revoke access at any time from your Google account settings.

Limited Use. WeMachines’s use and transfer of information received from Google APIs to any other app will adhere to the Google API Services User Data Policy, including the Limited Use requirements. In particular, we do not use Google user data - raw, aggregated, or derived - to develop, improve, or train generalized or foundational artificial intelligence or machine learning models, and the AI providers that process it are bound by their business-API commitments not to train their models on it either. Google user data is used solely to provide and improve the calendar and Gmail features you have chosen to use, is never sold, and is never transferred or used for advertising. No person at WeMachines reads your Gmail data unless you ask us to (for example, in a support request), it is needed for security purposes such as investigating abuse, or the law requires it.

Outlook

Connecting Outlook is optional, is never requested at sign-in, and connecting it never becomes a way to log in. It works with a Microsoft 365 work or school account or a personal Outlook.com account. If you connect it in Integrations → Outlook, Microsoft asks you to allow Mail.ReadWrite, which lets us search and read your email and save a draft in your mailbox, and Mail.Send, which lets us send email from it, together with User.Read (your name and address, so we know whose mailbox it is) and offline_access (so the connection lasts beyond an hour). The resulting access and refresh tokens are stored encrypted at rest and are readable only by our servers.

We use that access only when you ask your personal AI teammate about your email, to search, read, draft and send what you asked for. An email goes out on its own only to members of your workspace, to you, or to an address you wrote in your request; an email to anyone else is saved as a draft and sent only when you press Send on it yourself, in WeMachines or in Outlook. The relevant messages are sent to our AI providers to generate the response, as described in “AI processing” above, and are not used to train AI models. Only your personal AI teammate can use your connection, only in the teams you turn it on in. We do not copy your mailbox into our database, we do not read it in the background or on a schedule, and no other member of your workspace can use your connection. Disconnecting it from every team in Integrations → Outlook deletes the stored tokens; you can also remove WeMachines’s access at any time from your Microsoft account’s app permissions (for a work or school account, your organization’s administrator can too).

7. How we share information

We do not sell your personal information. We share information only as needed to run the Service: with the members of your workspace; with the service providers (sub-processors) listed below; in connection with a merger, acquisition, or sale of assets (with notice where required); and when required by law or to protect rights, safety, and the integrity of the Service.

Sub-processors

We rely on the following providers to operate the Service. They process data on our behalf:

ProviderPurpose
SupabaseAuthentication, database, file storage, and real-time sync
CloudflareHosting, serverless functions, cloud containers for the coding/voice machines, real-time media, and bot prevention; and WeMachines' built-in AI - runs DeepSeek's V4.1 Flash model, the default model, through Workers AI, keeping no data: summaries and other everyday AI features, and replies from AI teammates on the Efficient level when their own model is unavailable
FingerprintBot and fraud prevention at sign-up - recognises the device a new account is created from
OpenRouterAI gateway - receives every AI request and passes it to an endpoint that runs the model, pinned to endpoints that keep no data (zero data retention) wherever the model has them; and WeMachines' built-in AI for AI teammates on the Efficient and Balanced levels - runs a model built on DeepSeek's V4.1 Flash (Efficient) and Moonshot AI's Kimi (Balanced), each on an independent inference provider, keeping no data
Amazon Web ServicesWeMachines' built-in AI - runs Anthropic's Claude models through Amazon Bedrock, keeping no data: pull request descriptions, and repository studies, which read the source files of a team's connected repositories to describe what each one holds and how they connect, when an admin starts a study
Google CloudWeMachines' built-in AI - runs Google's Gemini models through Vertex AI, keeping no data, for quick, high-volume tasks such as sorting new chat messages and for everyday AI features while the default model is unavailable, and Anthropic's Claude models, as Amazon Bedrock does
MicrosoftWeMachines' built-in AI - runs OpenAI's GPT models through Azure, keeping no data, for tasks AI teammates take on, app builds and other in-depth AI features, Microsoft's MAI Image model for the images AI teammates make, and OpenAI's embedding model for the notes AI teammates remember
Voice model hostsVoice calls with AI teammates - what is said on a call goes to DeepSeek's V4.1 Flash model, which OpenRouter runs on independent inference providers that keep no data, picking the fastest for each call from its list of zero-data-retention endpoints; none of it goes to DeepSeek's own service
Other AI model providersA model WeMachines moves a feature onto before this list names its host, and images made with an image model other than the default; content for those features goes to wherever that model runs, and this list is updated when it changes
LinkUpWeb research for AI teammates - receives the research questions, keeping no data (zero data retention)
ApifyGrowth data for AI teammates - receives what is looked up (account names, search words or links on public platforms) and collects the public data it asks for
BrowserbaseThe web browser AI teammates use - the pages they open, what they type into them and the sites they log in to; sessions are not recorded or logged
GitHubSource-control integration (repository access, branches, pull requests), and the build servers that deploy WeMachines and take the nightly backup of its database and file storage
SlackSlack app - messages in the public channels machines learn from, who is who in your Slack, and the replies and tasks it posts back, where a workspace connects it
DiscordDiscord app - messages in the channels machines learn from, the linked accounts of the members it answers, and the replies it posts back, where a workspace connects a server
TelegramTelegram bot - messages in the groups a workspace connects, and its replies to people who write to it
Microsoft TeamsMicrosoft Teams app - messages in the channels of the team a workspace connects, and the notices it posts there
GoogleSign-in with Google, and Google Calendar and Gmail - the events on the calendar and the email you connect, where you connect them
LiveKitReal-time audio and video for huddles and the voice assistant, and the speech-to-text and text-to-speech behind the voice assistant and huddle transcription (through LiveKit Inference, which uses Google to hear and Cartesia to speak)
StripePayment processing, billing, and tax calculation
MailgunTransactional and product email delivery - including emails that name teammates and carry a task's title or a line of a message, such as the first-day and missed-mentions emails
LoopsProduct update emails - your email address, first name, account identifier and whether you are subscribed, plus product events such as a workspace being created, a tool connected or a plan started; you can turn the emails off in Settings → Profile or from any of them
PostHogProduct analytics and diagnostics (no session replay)
ApolloIdentifying the companies that visit wemachines.com, only if you accept analytics - the pages you view, your IP address and campaign tags, and, through its partner LiveIntent, a hashed email address where LiveIntent can match your browser
lemlistIdentifying the companies that visit wemachines.com, only if you accept analytics - the pages you view, time on each page, clicks, your IP address and browser type
FigmaFigma integration - the files a workspace connects, where it connects one
GiphyGIF search in chat - the search terms, and a one-way coded identifier for each person that Giphy cannot trace back to them; GIFs in chat are loaded from Giphy
ApplePush notifications to the desktop and mobile apps - the notification's text, such as a task or page title, where you allow notifications

This list may change as the Service evolves; we will keep it up to date here. Some providers process data in the United States and other countries.

8. Data retention

We keep workspace content for as long as the workspace is active, because it is the workspace. Chat messages, huddles, completed tasks, documents, pages and everything else stay until someone deletes them or the workspace is deleted. Three things move into a grey zone after 14 days, by design and visibly in the product, where they are shown greyed out and can still be searched: chat messages, counted from when a thread was last active; huddles, counted from when the call started; and completed tasks, counted from when the task was moved into a done status, so moving it back out starts the window again. The rest is kept only as long as it has a purpose:

DataKept for
Chat messages, huddles, completed tasksLife of the workspace (in the grey zone after 14 days)
What AI teammates remember: their notes and the words they came fromAs long as the messages they came from: deleting a message deletes its words and every note learned from it, and any note can be deleted on its own
Slack messages from a connected Slack90 days. What AI teammates learned from them stays until it is deleted, learning from Slack is switched off, or Slack is disconnected
Discord messages from a connected server90 days. What AI teammates learned from them stays until it is deleted, learning from that channel or server is switched off, or the server is disconnected
Telegram messages from a connected group90 days. What AI teammates learned from them stays until it is deleted, learning from that group is switched off, the group is disconnected, or the person who wrote them sends /forget
Microsoft Teams messages from a connected team90 days. What AI teammates learned from them stays until it is deleted, learning from that team is switched off, the team is disconnected, or the person who wrote them sends "forget" to the bot
Other workspace content and account profileLife of the workspace or account
Diagnostic and request logs30 days
Product analytics12 months
Support correspondence3 years from the last message
Billing and tax records7 years, as tax law requires
BackupsOverwritten on a rolling 30-day cycle

When you delete content, your account, or your workspace, we delete or anonymize the associated data across our systems and instruct our sub-processors to do the same, within 30 days. The exceptions are the records above that we are required to keep, principally billing and tax records, and anything we must retain to resolve a dispute or enforce our agreements. Deleted data may persist in backups until that cycle overwrites them, and is not restored to live systems in the meantime.

9. Security

We take technical and organizational measures to protect your information, including encryption in transit, encryption of sensitive credentials (such as connected-account tokens and the secrets a workspace stores for its machines and apps) at rest, and database-level access controls that isolate each workspace’s data. Our architecture, tenant-isolation model, and vulnerability-reporting process are described in detail on our Security page.

No method of transmission or storage is completely secure. A workspace admin can export the workspace’s full contents at any time, and we recommend keeping your own copy of anything critical.

10. Your rights

Depending on where you live, you may have rights to access, correct, delete, export, or restrict the processing of your personal information, and to object to certain processing or withdraw consent. You can update much of your profile and content directly in the app. A workspace administrator can download the workspace's full contents at any time from Settings → Danger Zone → Export workspace, without asking us - no plan requirement and no request form. You can delete your account yourself from Settings → Danger Zone → Delete account in the app, which also deletes every workspace where you are the only member. To delete a workspace you share, or to exercise any of these rights, contact us at contact@wemachines.com and we will action the request as required by applicable law. If a workspace administrator controls your workspace, some requests may need to be directed to them.

If you are in the European Economic Area, the United Kingdom, or Switzerland, you also have the right to lodge a complaint with your local data protection supervisory authority.

11. International transfers

We and our providers operate globally, and your information may be processed in countries other than your own, including the United States. Where personal data leaves the European Economic Area, the United Kingdom, or Switzerland, we rely on the European Commission's Standard Contractual Clauses (with the UK Addendum where the UK GDPR applies), agreed with each provider that receives it, and on an adequacy decision where one covers the destination. Several of our providers are additionally certified under the EU-US Data Privacy Framework. You can request details of the safeguards that apply to a specific provider at contact@wemachines.com.

12. Children

The Service is not directed to children, and you must be at least 16 years old (or the age of digital consent in your jurisdiction, if higher) to use it. We do not knowingly collect personal information from children below that age.

13. Billing & refunds

Paid-plan payments are processed by our payment provider, Stripe, which handles your payment and billing information. WeMachines, Inc. is the seller of record, and we do not store full payment card details. Our billing, cancellation, and refund terms - including that subscriptions can be cancelled anytime, remain active until the end of the paid period, and that payments are non-refundable except where required by law - are set out in our Terms of Service.

14. Changes to this policy

We may update this Privacy Policy from time to time. If we make material changes, we will update the “Last updated” date above and, where appropriate, provide additional notice. Your continued use of the Service after changes take effect constitutes acceptance of the updated policy.

15. Who is responsible for your data

The Service is operated by WeMachines, Inc., a Delaware corporation with its registered office at 2810 N Church St STE 89498, Wilmington, DE 19802, United States.

Which role we play depends on the data, and the distinction matters if you are exercising a right:

  • We are the controller for the data we hold to run WeMachines as a business: your account and profile, billing records, support correspondence, product analytics, and the diagnostic logs described above. Requests about that data come to us.
  • We are a processor for the content inside a workspace - the messages, tasks, documents, comments and files your team creates, and the personal data they may contain. The organisation whose workspace it is controls that content; we process it on their instructions to provide the Service, and our Terms of Service incorporate a Data Processing Agreement setting out that relationship. If your workspace belongs to your employer or client, a request about workspace content is best directed to them, and we will support them in answering it.

That Data Processing Agreement already applies to your workspace, with nothing to sign; a copy, countersigned if you need it, comes from contact@wemachines.com. Questions about your privacy or this policy, and any of the requests described in “Your rights” above, can be sent to the same address.