This Privacy Policy describes how Androcrew ("we," "us," or "our") collects, uses, discloses, and protects your information when you access or use the Androcrew platform at chat.androcrew.com and any related services, websites, mobile applications, APIs, and integrations (collectively, the "Service"). Androcrew is an AI-powered personal assistant platform that uses generative AI and large language models ("LLMs") to perform tasks on your behalf.
Privacy at a glance
- We collect only what is needed to operate the Service.
- We do not sell your personal information.
- Androcrew does not use your conversations, files, or connected account data to train AI models. The AI model providers that answer your requests apply their own terms, see Section 4.2.
- For financial connections (Plaid), we store an encrypted access token and basic details about each linked account. We do not store your bank transactions or balances. We fetch them when you ask and discard them after the response is generated.
- We do not record any calls. Call audio is never saved. Speech is converted to text during the call, and a text version of the conversation is kept with the call details (see Section 7).
- When Andi checks in without being asked, the background review is read only. Andi does not send a message or change your calendar without your approval of the exact action.
- You can delete your chats, files, Library items, memories and connected accounts in the Service at any time, and you can ask us to delete your entire account.
1. Information We Collect
1.1 Account Information
When you create an account, we collect:
- Email address
- Name (as provided via your identity provider)
- Authentication credentials managed through our identity provider (Keycloak); we do not see or store your password directly
- Optional profile information you choose to provide (preferences, locale, timezone)
1.2 Conversation Data
When you use the Service, we store:
- Messages you send and responses generated by AI, including the text of what you say in voice chat
- Chat session metadata (titles, timestamps, tags, folders)
- Summaries of long conversations, which the Service creates so Andi can keep track of earlier context
- Files you upload to conversations or knowledge bases
- Items saved in your Library (see Section 10)
- Notes and personal context you save within the platform
- Vector embeddings (numerical representations) of your content used for search and retrieval
1.3 Connected Account Data
If you connect third-party accounts (e.g., Google, an email account, Plaid, or a connector), we may access and store:
- OAuth access and refresh tokens, encrypted at rest, used to maintain your authorized connection
- For email accounts you connect by username and password (IMAP or POP3), the password, encrypted at rest, and the server settings and username needed to connect
- The minimum profile metadata needed to identify the connection (e.g., the email address of the connected mailbox)
- Data returned from the connected service in the course of fulfilling a specific request you make (see Section 4 for how this data flows through AI processing). Email messages Andi opens for you during a chat are held in a short-lived cache for up to one hour so Andi can refer back to them, and are not otherwise stored by the email feature.
We request only the OAuth scopes needed to provide the features you use. You can disconnect a Google, email or bank account at any time. Disconnecting deletes the sign-in credentials we stored for that account, and for Google we also revoke our access at Google. Connectors are covered in Section 12.
1.4 Billing Information
Payment processing is handled by Stripe. We store:
- Subscription tier and status
- Stripe customer and subscription identifiers
- Last four digits and brand of your payment method (for display only)
- Invoice history and billing address (as provided to Stripe)
We never store full credit card numbers, CVVs, or raw payment credentials.
1.5 Usage Data
- Token consumption, usage and task counts per billing cycle
- API key usage metadata (API keys themselves are stored only as hashes)
- Standing instructions and watches you set up with Andi, and their run history
- Proactive Andi settings and check-in history (see Section 9)
- Feature interaction events (e.g., which crew was selected, which tools were called)
1.6 Automatically Collected Data
- Request logs (IP address, user agent, request timestamps, request IDs)
- Error and performance logs for service reliability
- Approximate location derived from IP address (for security and regional routing)
2. How We Use Your Information
| Purpose | Data used |
|---|---|
| Provide AI-powered responses | Conversation messages, uploaded files, connected account data fetched for the request |
| Maintain chat history, your Library and personalization | Messages, session metadata, Library items, saved memory entries |
| Run the standing instructions and watches you set up | Instruction and watch text, user context, connected account access needed for the instruction |
| Decide whether to check in with you and prepare suggestions (Proactive Andi) | The sources listed in Section 9 |
| Operate the browser agent and connectors you ask Andi to use | Your instructions, page content, data sent to connectors you add |
| Process payments and manage subscriptions | Billing information via Stripe |
| Enforce usage limits and prevent abuse | Token consumption, usage, task counts, request logs |
| Operate, debug, and improve service reliability | Error logs, performance metrics, AI observability traces |
| Send service notifications | Email address and browser push subscription (usage warnings, billing updates, security alerts, Andi's updates) |
| Comply with legal obligations and enforce our Terms | Account, billing, and log data as required |
We do not use your conversations, files, or connected account data for advertising, profiling for marketing purposes, or to train AI models of our own.
3. Legal Bases for Processing (EEA and UK)
If you are located in the European Economic Area or the United Kingdom, we process your personal data under the following legal bases:
- Contract. To provide the Service you have requested.
- Legitimate interests. To operate, secure, and improve the Service, prevent fraud and abuse, and communicate with you about the Service. We balance these interests against your privacy rights.
- Consent. For optional features such as connecting third-party accounts, telephony features, and certain cookies. You can withdraw consent at any time.
- Legal obligation. To comply with applicable laws and respond to lawful requests.
4. AI and Large Language Models
Important. Your messages, attached content, and connected account data fetched for a request are sent to third-party AI model providers to generate responses. This is core to how the Service works.
4.1 How AI Processing Works
- Your messages, plus any context the assistant determines is relevant (recent chat history, retrieved knowledge base snippets, tool results, fetched calendar events or emails, etc.), are assembled into a prompt and sent to one or more third-party LLM providers.
- The same happens when Andi works in the background, for example when Proactive Andi reviews what is new (see Section 9) or when a standing instruction runs.
- Responses are streamed back to you and stored as part of the conversation.
- Uploaded files, knowledge base documents and your messages are processed to generate vector embeddings for search and retrieval.
- Text that comes from emails, websites and other outside sources is treated as untrusted, so it cannot direct Andi to take actions.
4.2 AI Model Providers
We route LLM requests through OpenRouter, which forwards them to upstream providers including Anthropic, Google, OpenAI, xAI, and others. Each provider processes the requests it receives under its own terms and privacy policy.
OpenRouter and the providers it forwards to may be able to keep your requests and use them under their own terms, for example to improve their services. We do not control that, so please assume they can. Read OpenRouter's privacy policy if you want the details, and avoid sending through Andi anything you would not be comfortable sharing with an AI model provider.
4.3 Training and Your Data
Androcrew does not use your conversations, prompts, completions, files, embeddings, or connected account data to train or fine-tune any AI model of our own, and we do not sell this content. What OpenRouter and the AI model providers do with the requests they receive is covered by their own terms, as described in Section 4.2. They may be able to keep and use that content, so we cannot promise that they will not.
4.4 AI Limitations
- AI-generated responses may be inaccurate, incomplete, biased, or otherwise inappropriate.
- The Service does not provide professional advice (legal, medical, financial, tax, or otherwise).
- You should independently verify any important information before relying on it.
4.5 Observability
We use Langfuse, which we host ourselves, to monitor AI interactions for quality, debugging, latency, and cost. This can include logging prompts, responses, tool calls, and token usage. Access is restricted to authorized members of our operations team, and traces are deleted on the schedule in Section 17.
5. Memory and Personalization
The Service can remember facts and preferences you choose to share (or that the assistant infers from your messages) to provide more personalized responses across sessions. Memory entries are stored in your account, are visible to you, and can be edited, exported or deleted at any time. Memories saved without you asking expire after 180 days. When a memory is deleted, it is kept in an archive for up to 90 days and then permanently deleted. Memory is used only to personalize your experience and is not shared across users or used to train models.
6. Financial Account Linking (Plaid)
We do not store your bank transaction data. If you link a financial account via Plaid, we store an encrypted Plaid access token tied to your account, along with basic details about each linked account. When you make a request that requires financial data (for example, asking about your spending or your subscriptions), we use that token to fetch the necessary data from Plaid in real time, process it to answer your request, and discard it from working memory once the response is generated. Bank transaction details, balances, and full account numbers are not persisted in our databases.
6.1 What We Store
- An encrypted Plaid access token and item identifier per linked institution.
- Display details about the link so you can identify and manage your connections. These are the institution name and, for each account, its name, type, and the last few digits of the account number as provided by Plaid.
- A position marker that lets us ask Plaid only for changes since the last request.
- Budget settings you create, such as budget goals, custom categories, merchant rules and alert preferences, and a history of the budget alerts we have sent you.
6.2 What We Do Not Store
- Bank login credentials (Plaid handles authentication; we never see your credentials).
- Full account or routing numbers.
- Transaction histories, transaction-level metadata, balances, holdings, or liabilities.
Answers that Andi gives you based on your financial data, such as a spending summary, are saved in your chat like any other response.
6.3 Data Flow
For each financial request, we retrieve your access token, fetch the data from Plaid, analyze it in memory, return the answer to you, and then discard the fetched data. Anything that briefly appears in our AI observability traces is access restricted and is deleted on the schedule in Section 17.
6.4 Disconnecting
You can disconnect any linked institution at any time from the Service, which deletes the stored access token. You can also revoke access directly through Plaid Portal.
7. Voice and Telephony Features
We do not record any calls. When the assistant places a phone call on your behalf, or when you talk with Andi using voice chat, the audio is converted to text while the call or voice session is in progress. The audio is not saved.
7.1 Phone Calls
When a crew places a call for you, for example to book an appointment, the call is carried by our telephony provider, the other party's speech is converted to text by our speech to text provider, and the assistant's replies are spoken by our text to speech provider (see Section 13). For each call we keep:
- Call details, such as the phone number called, the goal and instructions for the call, timestamps, duration, and the outcome.
- A text version of the conversation produced from the speech on the call, stored with the call details.
- Information taken from the call to complete your task (for example, the appointment times a business offered), and Andi's report of the call in your chat.
- Short, truncated excerpts of call text in our system logs, used for debugging and support.
7.2 Voice Chat in the App
When you use voice chat, your microphone audio is streamed to our servers and to our speech to text provider so it can be converted to text, and Andi's replies are spoken by our text to speech provider. The text of what you say and Andi's replies are saved as messages in your chat. The audio is not saved.
7.3 What We Do Not Retain
- Audio recordings of voice chat or telephony calls.
7.4 Two-Party Consent and Disclosure
Some jurisdictions require all parties to a call to consent to recording or transcription. Because the Service converts speech on calls to text and keeps that text, you are responsible for providing any required disclosure to the other party and for using the feature in compliance with applicable laws. When placing calls on your behalf, Andi says whose assistant is calling and confirms that the call comes from an AI assistant if asked.
8. Google User Data
Where the Service integrates with Google APIs (Gmail, Calendar, Contacts), our use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. We use Google user data only to provide and improve user-facing features you have requested, including the Proactive Andi check-ins described in Section 9. We do not transfer Google user data to third parties except as necessary to provide or improve those user-facing features, to comply with applicable law, or as part of a merger, acquisition, or sale of assets with notice to users. We do not use Google user data for serving advertisements or for training generalized AI models.
9. Proactive Andi (Background Check-ins)
Andi can look out for you in the background and check in with you in your main chat when something seems worth your time. Proactive Andi is available on every plan and is turned on by default. You can turn it down or off at any time (see 9.6).
9.1 What Andi Reviews
To decide whether to check in, Andi periodically reviews a limited set of information. Which sources are included depends on your plan and on the accounts you have connected.
- Your chats. A summary and the most recent messages of your main chat, the titles of your recent chats, and earlier check-ins.
- Your standing instructions and watches, and how they last turned out.
- Saved memories, such as your preferences, goals and facts you have shared, and your feedback on earlier check-ins.
- Your calendar, if connected. Events in the next 48 hours, with their title, time, location, status and number of guests.
- Your email, if connected, on the Personal plan and above. For Gmail, Andi reads the sender, subject, date and the short preview that Gmail provides, not the full message. For email accounts connected by username and password (IMAP), the message is downloaded so that a short preview can be made from it.
- Information your crews keep, on the Personal plan and above. Summaries of items such as complaint cases, bill negotiations, job applications and job watches, gift occasions, party and social events, home maintenance reminders, and budget goals. Budget summaries do not include transactions or amounts.
- Your Library, on the Personal Pro plan and above.
Once a week, Andi also reviews your recent main chat messages to learn which kinds of check-ins you find useful. This information is processed by AI model providers as described in Section 4. Text from emails, calendars and websites is treated as untrusted and cannot direct Andi to act.
9.2 Background Work Is Read Only
When Andi works in the background, the work runs in a read only mode. In this mode the Service is designed to block sending, replying, deleting, booking, calling, buying, and changing your calendar or other accounts. The browser agent and connectors you add are not used in background check-ins.
9.3 Nothing Happens Without Your Approval
Andi can prepare an action for you ahead of time, such as a draft reply or a calendar change. You see the exact action on an approval card, and nothing is carried out until you approve it. When you approve, our systems check that the approval matches that exact action before it runs, so text in an email or on a website cannot fake an approval. Prepared actions that you do not approve expire after 72 hours.
Preparing work in advance uses your plan's usage, so Andi asks your permission before doing so. This preparation is not available on the Free plan. Standing instructions you create also use your plan's usage when they run.
9.4 What We Store
- Check-ins Andi posts, which are ordinary messages in your main chat.
- A record of each check-in, suggestion and prepared action, including the text of any prepared draft (for example its recipients, subject and body) and your response to it. These are deleted after 180 days.
- Your Proactive Andi settings, such as your level, time zone and quiet hours, and your answer about preparation work.
- Short fingerprints and position markers for each source, so Andi can tell what is new without storing the content itself, and counts of how often each topic has come up.
- Your feedback, such as "useful," "not useful," or "stop suggesting that," which is also saved as a memory.
When Andi reviews your email in the background, an email's sender, subject and preview are held only until the related check-in is posted, dropped or expires. After that they remain only in the text of a check-in Andi posted to you or a draft Andi prepared for you.
9.5 Logs
We do not write the content Andi reviews to our system logs. For Proactive Andi, logs record identifiers, counts and outcomes only, not subjects, previews, names, addresses or amounts. For the steps that decide whether to check in, our observability tools record usage information such as model, token counts and timing, not the content.
9.6 Limits and Your Controls
- Quiet hours. By default Andi does not check in between 9 p.m. and 8 a.m. in your time zone. During quiet hours, an urgent update may be posted to your chat without a notification, at most once every 12 hours.
- Daily limits. The number of check-ins each day is limited, and small updates are bundled into one daily update.
- Turning it down or off. Tell Andi in any chat to check in less, more, or not at all, to stop suggesting a topic, or to change your quiet hours.
- Notifications. Push notifications are delivered through your browser's push service in encrypted form and can be turned off in settings. Email updates have their own setting with the choices On, Only when I've been away, and Off.
10. The Library
The Library is a place in the Service where the lists, boards and trackers Andi builds for you are saved, together with your images and files.
10.1 What We Store
- Pages Andi builds for you, such as checklists, shopping lists and trackers, including the page itself, its title and summary, and a link to the chat that created it.
- The data you save on a page, such as checkmarks and notes.
- Images and files, including files you upload and documents your crews create for you, such as a tailored resume.
- How you organize your Library, such as collections and pinned items.
Library items are stored in our database on Google Cloud Platform, and files are stored in Google Cloud Storage. Pages are created with the help of the AI model providers described in Section 4.
10.2 Live Budget Numbers
A budget tracker pulls fresh numbers from your connected accounts each time you open it. Those numbers are not saved in your Library.
10.3 How Library Pages Are Protected
- Your Library is private to your account and has no public sharing.
- Every page runs in a sandbox, kept apart from the rest of the Service. A page cannot connect to the internet on its own and cannot see your login or your other items.
- A page can save only its own data, and our systems check every save to confirm that it belongs to you and is a reasonable size.
10.4 Deleting Library Items
You can delete any Library item at any time. Deleting a file also deletes the stored file. Library items are not deleted automatically. We keep a technical log of Library page builds that contains identifiers and usage counts but no page content, and that log is not removed when you delete an item.
11. The Browser Agent
When you ask, Andi can open and use websites for you in a cloud browser that we operate.
11.1 How It Works
- The browser runs on our cloud infrastructure, in a separate, sandboxed environment for each user. It is not the browser on your own device, and it cannot see your device's cookies, saved passwords or files.
- Andi uses the browser agent only during a conversation with you, not in background check-ins.
- You can watch the browser live and take control of it. The live view is passed through to you and is not stored.
- To decide what to do next, Andi sends the content of the pages visited to the AI model providers described in Section 4. Page content is treated as untrusted, so a website cannot direct Andi to take actions. Andi's replies about what was done are saved in your chat.
- Each user's cloud browser presents a consistent browser identity, so websites may recognize visits from it as coming from the same browser.
- Browser sessions close after a period of inactivity, and the browser profile is discarded after each session.
11.2 What It Will and Will Not Do
- Andi does not type passwords, one-time codes or payment card numbers. When a site asks for them, Andi hands control of the browser to you so you can enter them yourself.
- When a site shows a human check, such as a CAPTCHA, Andi hands control to you to complete it.
- Andi asks for your approval before buying something, posting, sending a message, deleting something, changing account settings, signing a document, or submitting a form that contains sensitive information or payment details.
- Andi may fill in and submit everyday forms that do not contain sensitive information, such as a booking or a contact request, as part of completing the task you gave Andi.
11.3 Saved Site Sessions
If you sign in to a website while Andi's browser is open, the site's session data (its cookies and site storage) can be saved so that Andi stays signed in to that site next time. Saved logins are locked with a passkey that you create on our separate vault page at vault.androcrew.com.
- Setting up your passkey. The first time a login can be saved, you are offered the chance to set up your passkey, and the login is saved only after you do. Your passkey stays on your device or in your password manager, and we never receive it. Some devices and password managers do not support the passkey feature this needs, and on those Andi cannot save logins with a passkey.
- What we store. The saved session data is encrypted so that it can be opened only with your passkey or your recovery code, and it is stored in Google Cloud Storage. We also store a locked copy of your vault key for each passkey you add and one for your recovery code, and only that passkey or that code can open it. While your logins are stored, no one at Androcrew can open them, because we do not have your passkey or your recovery code.
- Using a saved login. When Andi's browser needs your saved logins, you unlock them with your passkey on vault.androcrew.com and choose whether to restore all of your saved sites or only the site Andi is starting on. The keys for those sites are sent in encrypted form to your own browser session and are used only for that session. If you do not unlock, Andi uses the browser without those logins. Once a site is restored, Andi uses it in our cloud browser as described in 11.1.
- Your recovery code. When you set up your passkey, you are shown a recovery code. Keep it somewhere safe and only ever type it on vault.androcrew.com. Andi will never ask for it in chat or anywhere else. You can also add passkeys from your other devices.
- If you lose your passkey. You can unlock with your recovery code instead, and we recommend replacing your vault keys afterwards. If you have lost both your passkeys and your recovery code, we cannot open your saved logins for you. You can forget them, or reset your vault. A reset needs a fresh sign in to your Androcrew account with a second verification step, takes effect after 72 hours, and can be cancelled during that time. When you set up a new passkey after a reset, your old saved logins are deleted and you sign in to those sites again.
- Deleting saved logins. You can forget the saved login for one site, or for all sites, at any time on the Connected Sites page in the app. Saved logins do not expire automatically.
- Logins saved before passkeys. When we moved saved logins to passkey protection, we deleted every login saved under the earlier method. To let Andi use one of those sites again, sign in to it again in Andi's browser.
12. Connectors You Add
The Service lets you add connectors, which link Andi to an outside server or service that you choose, either from our directory or by entering the server's address yourself.
- What we store. The connector's address, the list of tools it offers and your settings for them, and any "Allow always" choices you make. Sign-in tokens for the connector, and any client secret you provide, are stored encrypted.
- What the connector receives. When Andi uses a connector, the connector's server receives the information Andi sends to it to carry out the request, which can include information from your conversation, along with your access token for that connector. The connector's response is sent to our AI model providers so Andi can use it, and it is treated as untrusted.
- Your approval. Andi asks for your approval before each use of a connector unless you have chosen "Allow always" for that tool. New or changed tools stay off until you review them.
- Your responsibility. Connector servers are operated by third parties, or by you, not by Androcrew. Each connector's operator receives and handles the data sent to it under its own terms and privacy policy. You are responsible for the connectors you add and for deciding what data to send to them.
- Removing a connector. You can remove a connector at any time, which deletes its stored settings.
13. Third Party Services
The Service integrates with the following third-party services:
| Service | Purpose | Data shared |
|---|---|---|
| OpenRouter and upstream LLM providers | AI model routing and responses | Prompts and contextual content needed to generate the response |
| Google APIs | Gmail, Calendar, Contacts | OAuth tokens; data accessed per your requests and for Proactive Andi check-ins |
| Plaid | Financial account linking | Bank account access tokens; transactional data fetched on demand and not persisted |
| Stripe | Payment processing | Billing and payment information |
| SendGrid | Transactional email | Email address, notification content, and for party invitations you send, your guests' email addresses |
| Twilio | Placing phone calls | Phone numbers and call audio while the call is in progress |
| Deepgram | Speech to text for calls and voice chat | Call audio and voice chat audio while the call or session is in progress |
| ElevenLabs | Text to speech for calls and voice chat | The text the assistant speaks |
| Firecrawl | Reading web pages for research, job searches and other crews | Addresses of the pages to read, which can include your search terms |
| Serper | Web search | Search queries |
| SerpApi | Flight search | Flight search details, such as places and dates |
| Google Places API | Place and location search | Place and location queries |
| Google Gemini API | Image generation | Image prompts and images you ask to edit |
| Cloudflare Turnstile | Spam protection on forms | A challenge token, IP address and browser information |
| Browser push services | Delivering push notifications you turn on | Notification content, encrypted so the push service cannot read it |
| Connectors you add | Tools you connect to Andi (see Section 12) | The information Andi sends to the connector at your request, and your access token for it. You are responsible for the connectors you add |
| Websites the browser agent visits | Completing tasks you ask for (see Section 11) | The information Andi enters on the site for you, and standard browser information |
| Keycloak (self-hosted) | Authentication and SSO | Account and authentication data |
| Google Cloud Platform | Cloud infrastructure | All hosted data (under our control) |
Each third-party service is governed by its own privacy policy. We encourage you to review them.
14. How We Share Information
We share personal information only as described in this policy:
- Service providers and processors who help us operate the Service (cloud hosting, payment processing, AI model routing, email delivery, telephony, speech services, search) under terms that restrict their use of your information.
- Connected services you authorize, when you direct the assistant to send an email, create a calendar event, place a call, use a website, use a connector, etc.
- Chat share links you create. If you create a share link for a chat, anyone with the link can view that chat.
- Legal and safety reasons, where we believe disclosure is reasonably necessary to comply with law, enforce our Terms, or protect the rights, safety, or property of users or others.
- Business transfers, in connection with a merger, acquisition, financing, or sale of assets, with appropriate notice to you.
- With your explicit consent for any other purpose.
We do not sell your personal information, and we do not "share" it for cross-context behavioral advertising as those terms are defined under the CCPA and CPRA.
15. Cookies and Tracking Technologies
We use a small number of strictly necessary and functional cookies and similar technologies (such as browser local storage) to:
- Keep you signed in (session and authentication tokens).
- Remember your UI preferences (theme, language, layout).
- Maintain CSRF protection and other security controls.
- Remember the campaign or link that referred you (see 15.2).
We do not use third-party advertising or cross-site tracking cookies. Where required by law, we will request your consent for non-essential cookies.
15.1 Analytics on Our Marketing Website
We measure traffic to our public marketing website at androcrew.com using Umami, an analytics tool that we host ourselves on our own infrastructure. The analytics script is served from our own domain and the resulting data is stored in our own database. No third-party analytics or advertising provider receives it.
This analytics data is limited to what is needed to understand how the website performs. Umami keeps an anonymous session and the pages that session visited, along with the site or campaign that referred you and general device, browser, and screen information. It holds nothing that identifies you as a person. Umami does not set advertising or cross-site tracking cookies, does not create a persistent identifier that follows you to other websites, and does not link marketing website visits to your Androcrew account.
15.2 Campaign Attribution
When you reach our marketing website or the Service through an advertisement, an email, or a link we have published, that link may carry campaign parameters (commonly called UTM parameters, such as utm_source and utm_campaign) and an advertising platform click identifier (for example gclid from Google Ads or fbclid from Meta). We also note the address of the page that referred you and the first page you visited.
These values are held in your browser's local storage for up to 30 days. If you create an account within that period, we associate a single attribution entry with your account and retain it for the life of the account. If you do not create an account, the values expire in your browser and are never associated with you.
We use attribution data solely to measure the effectiveness of our own marketing, for example to understand which campaigns introduce people to Androcrew. We do not transmit it to advertising platforms, we do not use it to build an advertising profile or to target advertising to you, and, consistent with Section 14, we do not sell or share it.
16. Data Storage and Security
16.1 Infrastructure
- Data is stored on Google Cloud Platform (GCP) infrastructure in the United States.
- Databases use PostgreSQL with encryption at rest.
- Files are stored in Google Cloud Storage with access controls.
- OAuth tokens, passwords for email accounts you connect, bank connection tokens, saved browser sessions and similar secrets are encrypted before storage. API keys are stored only as hashes.
- Traffic between your device and the Service, and between the Service and the outside providers listed in Section 13, is encrypted in transit using TLS (HTTPS).
16.2 Access Controls
- User data is isolated at the application level, so you can access only your own data.
- Administrative access is restricted and authenticated through SSO.
- We follow the principle of least privilege.
16.3 Your Responsibilities
No system is perfectly secure. You are responsible for keeping your account credentials and any API keys you generate confidential. Notify us immediately at support@androcrew.com if you suspect unauthorized access to your account.
17. Data Retention
| Data type | Retention period |
|---|---|
| Conversations and messages | Until you delete them or your account |
| Uploaded files and knowledge bases | Until you delete them or your account |
| Library items | Until you delete them or your account |
| Memory entries | Until you delete them or your account. Memories saved without you asking expire after 180 days. Deleted memories are archived for up to 90 days, then permanently deleted |
| Standing instructions and watches | Until you stop or delete them, or delete your account |
| Standing instruction and watch run history | 90 days |
| Proactive Andi check-in items and prepared drafts | 180 days |
| Automated action audit logs | 90 days |
| Saved browser site sessions | Until you delete them |
| Email messages opened during a chat | Up to 1 hour in a short-lived cache |
| Plaid transactional data | Not stored. Fetched on demand and discarded after the response |
| Voice and telephony audio | Not stored |
| Call details and the text version of call conversations | Kept with the call details. No automatic deletion period is set today. You can ask us to delete them at any time |
| Truncated call and voice text excerpts in logs | Retained with server logs (30 days) |
| AI observability traces (Langfuse) | 30 days |
| Server and request logs | 30 days |
| Campaign attribution data (UTM parameters, click identifiers, referring URL) | Until your account is deleted |
| Marketing website analytics (Umami; anonymous, not linked to accounts) | Up to 24 months |
| Billing records | As required by tax, accounting, and other applicable laws |
| Backups | Up to 35 days; deleted data is purged on the backup rotation cycle |
18. Your Rights and Choices
Subject to applicable law, you have the right to:
- Access the personal data we hold about you.
- Correct inaccurate or incomplete information.
- Delete your conversations, files, knowledge bases, Library items, memory entries, connected accounts, and your entire account.
- Disconnect third-party accounts at any time.
- Export a portable copy of your data.
- Object to or restrict certain processing.
- Withdraw consent where processing is based on consent.
- Lodge a complaint with your local data protection authority.
You can delete chats, files, Library items, memories, standing instructions and watches, saved site sessions, and connected accounts directly in the Service, and you can export your memories there. You can turn Proactive Andi down or off by telling Andi. For other requests, including full data exports and deletion of your entire account, email support@androcrew.com. We will verify your identity before acting on a request and will respond within the timeframes required by applicable law (generally within 30 days, extendable as the law permits).
19. California Privacy Rights (CCPA and CPRA)
If you are a California resident, the California Consumer Privacy Act, as amended by the California Privacy Rights Act, gives you specific rights regarding your personal information.
19.1 Categories Collected
In the past 12 months we have collected the following categories of personal information. These are identifiers (name, email, IP address, advertising click identifiers); customer records (billing details via Stripe); commercial information (subscription and usage data); internet and network activity (request logs, feature usage); geolocation (approximate, derived from IP); audio information processed in real time but not stored (voice chat and calls); and inferences drawn from any of the above for product personalization. We collect sensitive personal information only when you provide it (for example, content you place in a conversation or document, or account credentials you enter to connect an email account) and we use it solely to provide the Service.
19.2 Sources, Purposes, and Recipients
We collect this information directly from you, from your devices, and from third-party services you authorize. We use it for the purposes described in Section 2 and disclose it only to the categories of recipients described in Section 14.
19.3 Your California Rights
- Right to know what personal information we collect, use, disclose, and (if applicable) sell or share.
- Right to delete personal information we have collected from you.
- Right to correct inaccurate personal information.
- Right to limit the use and disclosure of sensitive personal information.
- Right to opt out of the sale or sharing of personal information. We do not sell or share personal information as those terms are defined under the CCPA and CPRA, so there is nothing to opt out of.
- Right to non-discrimination for exercising your rights.
To exercise these rights, contact support@androcrew.com. You may designate an authorized agent to make a request on your behalf, subject to verification.
20. Sensitive Information and Acceptable Use
The Service is a general-purpose AI assistant and is not designed or certified for handling regulated categories of data such as protected health information (HIPAA), card data subject to PCI-DSS beyond what Stripe handles, government-classified information, or data subject to specialized export controls. Please do not submit this kind of information to the Service. You are responsible for the lawfulness of the content you submit and for ensuring that you have the right to share any third-party data (including content from connected accounts) with the Service.
21. Automated Decision Making
The Service uses AI to generate suggestions, drafts, and actions in response to your requests. We do not use this AI to make decisions about you that produce legal or similarly significant effects without your involvement. Actions that have real-world impact (sending emails, placing calls, scheduling events, using websites, moving money where supported) require your authorization, either through your request at the time, your approval of an action Andi has prepared, or a standing instruction or watch you have set up.
22. Children's Privacy
The Service is not intended for users under the age of 18, and we do not knowingly collect personal information from children. If you believe a child has provided us with personal data, please contact support@androcrew.com and we will delete it.
23. International Data Transfers
Androcrew is operated from the United States, and your data may be processed in the United States and other countries where our service providers operate. Where required by law (for example, transfers from the EEA, UK, or Switzerland), we rely on appropriate transfer mechanisms such as the European Commission's Standard Contractual Clauses, the UK International Data Transfer Addendum, and applicable adequacy decisions. By using the Service, you understand that your data may be transferred to jurisdictions that may not provide the same level of data protection as your home jurisdiction.
24. Security Incident Notification
If we become aware of a security incident that affects your personal information, we will notify you and any applicable regulators as required by law, without undue delay. Notice will describe the nature of the incident, the categories of data involved, the steps we have taken or will take, and recommended actions for you.
25. Changes to This Policy
We may update this Privacy Policy from time to time. We will notify you of material changes by posting the updated policy on this page and updating the date shown at the top of this page. For material changes, we will also provide additional notice (such as an in-app notice or email) before the changes take effect. Your continued use of the Service after the changes take effect constitutes acceptance of the updated policy.
26. Contact Us
If you have questions about this Privacy Policy, our data practices, or wish to exercise any of your rights, email us at support@androcrew.com. Please use the same address to report a security concern.