Privacy Policy

• The product is an AI system. Argentic Code works by sending what you ask to a third-party AI model that you choose, and returning what that model generates. AI output can be inaccurate or incomplete, so review it, especially code and commands, before you rely on it. Our Terms of Service cover this in full.
• Your code stays on your machine. Repositories, files, and your conversations with the agent are not stored on our servers. There is one exception, and you start it: a feedback or bug report you send from inside the app attaches that session’s transcript, because looking at it is the point of the report. The dialog says so before you send.
• Your conversations are yours. Apart from a report you send us yourself, our admin tooling has no surface that returns conversation content, and we never use your content to train AI models.
• What we do collect is operational metadata — sign-ins, message and token counts, model names, error signals, plan and billing state — so we can run the product, bill correctly, prevent abuse, and understand aggregate usage. Not the content of what you write.
• We don’t sell your personal information and we don’t use advertising trackers.

• Account. Email address, name, and an optional profile photo, so your plan and settings stay attached to you. Sign-in is operated by us directly. If you create an account with a password, we never store the password itself: we store only a one-way bcrypt hash of it, which cannot be reversed back into your password. If you sign in with Google, no password exists on the account at all and we keep only the identifier Google returns. We also record whether your email address has been confirmed.
• Two-factor authentication. If you turn it on, we store the setup key your authenticator app uses, encrypted at rest, plus one-way hashes of your recovery codes (never the codes themselves) and the time of the last code you used, so the same code cannot be reused. Turning two-factor off deletes all of it. We never send authentication codes by SMS and store no phone number.
• Billing. Payments are processed by Stripe. We store your plan, subscription status, renewal date, and Stripe customer/subscription identifiers — never your card number.
• Conversations. We do not store your conversations. They are held on your own device. A retired feature of the app did save conversations and their attachments to our servers; that content is served only to the signed-in account that created it, is being removed, and is deleted outright if you delete your account.
• Provider credentials. If you connect an AI provider account (for example ChatGPT), we store the OAuth tokens encrypted at rest. API keys you paste into the app are kept in your device’s operating-system keychain, not on our servers, and are forwarded only transiently with each request you make.
• Gmail connector (optional). If you connect Gmail, we store OAuth tokens encrypted at rest, limited to the scopes you grant. We use it only to send email you explicitly direct, and we store a delivery record (status and message identifier) — never your mailbox contents.
• Contact and newsletter. If you use a contact or support form on our sites (argenix.ai or argentic.argenix.ai) we receive the name, email, company, subject, and message you submit; the newsletter stores your email address until you unsubscribe.
• Feedback and bug reports. If you send feedback from inside the app, we receive what you wrote and, attached to it, a transcript of the session you were in: your messages, the agent’s replies, and which tools it ran. Before that transcript leaves your device the app redacts file paths, email addresses, and anything shaped like a key or a token, shortens each tool result to a brief preview rather than sending whole files, and caps the length of the whole attachment. Nothing is sent unless you press Submit. These reports are used to fix and improve the product, they are readable by our administrators (the one place our tooling can show conversation content, and only for reports people chose to send), and they are deleted when you delete your account.
• Support. Anything you include in a support request.

• Event records. Operational events with timestamps: sign-in and other account-security events, completed AI turns (model used, settings, token counts, and response lengths as character counts — never the text), checkout started, connector activity, feedback submitted (that a report was sent and from which app version, never a word of what it said), and app-session events.
• Failed sign-in attempts. To stop people guessing their way into accounts, we count consecutive failed sign-in attempts and pause sign-in for a short period once there are too many. We store a one-way hash of the email address that was tried (not the address itself), the number of recent failures, and when the pause ends. We never store the passwords that were attempted. The record is deleted as soon as you sign in successfully or reset your password, and in any case within a day of the last attempt. When a pause begins on an existing account we email that account’s owner to tell them.
• Signed-out sessions. When you sign out, we record an identifier for that one session until it would have expired anyway, so the sign-out cannot be undone by anyone holding a copy of it. The record identifies the session, not you or your device, and is deleted once it lapses.
• Daily activity. Which calendar days your account used the Services — a per-day flag used for active-user and retention statistics, nothing more.
• Usage counters. A monthly count of AI messages, used to enforce your plan’s allowance.
• Subscription grant notices. If we grant your account a subscription (for example a promotional or support comp), we record the grant itself (plan, duration, and when the in-app notice was shown to you) and, if you tap the optional thank-you button on that notice, the time of the tap. Our administrators see a count of thank-yous and which accounts recently sent one (name, email, plan, and time), nothing more.
• Team membership and invitations. If you own or join a Team subscription, we record which team you belong to, your role on it (owner or member), and when you joined. When a team owner invites someone, we store the email address they entered, who sent the invitation, and when it expires, so the invitation can be delivered and redeemed. An invitation is stored even if the person invited has no account with us; it is deleted when it is revoked, and invitations that are never accepted expire after seven days. Team owners can see the name and email of everyone on their team and any invitation they have sent. Members see the other people on the team. Our administrators see only counts: how many teams exist, how many seats are paid for and filled, and how many invitations are outstanding, never who is on which team.
• Error diagnostics. When something breaks in the desktop app, the app records the failure instead of interrupting you with it, and sends us a diagnostic report: the error’s own message and stack trace, where in the app it happened, the app version, and your operating system. Before the report leaves your device we remove the parts that could identify you: your username is stripped out of any file path, and email addresses and anything shaped like an access token are replaced. We do not send your conversations, your files, your code, or the contents of what you were doing. Identical failures are grouped together, so what our administrators see is one entry per fault with a count of how many accounts it has affected, not a separate record of your individual crash. Reports are tied to your account only so that count is accurate; deleting your account removes your reports from it. Groups are deleted 90 days after the last time the fault occurred.
• Device and network. IP address and browser or app user-agent, recorded with security-relevant events for abuse prevention and audit.
• No advertising trackers. We use no advertising cookies and no cross-site tracking on the sites or in the app. Our websites use Vercel Web Analytics, a cookie-less service that shows us aggregate page views (the page viewed, referrer, country, and device type); it stores nothing on your device and identifies a visit only by an anonymized hash that changes daily. The app itself contains no analytics trackers. Local storage on your device holds your preferences (such as theme). We set one functional cookie, which carries your sign-in session and is what lets signing in on either argenix.ai or argentic.argenix.ai sign you in on both. It is used only to keep you signed in, never for advertising or tracking, and is cleared when you sign out or delete your account. In the desktop app there is no such cookie: your session is stored encrypted in your operating system’s keychain.

Argentic Code works on your local workspace. Your repositories and files, your conversations with the agent, and pasted API keys live on your device and are not stored by us. When the agent needs a model response, the relevant content (for example a prompt containing a file snippet) is transmitted through our servers to the AI provider you selected to generate that response — in transit only; we do not store it.

There is one way a conversation ends up stored by us, and it only happens when you ask for it: sending a feedback or bug report from the app’s account menu attaches that session’s transcript, redacted and shortened as described in section 2. If you never send a report, nothing from your sessions is kept on our servers.

• to provide, operate, and maintain the Services and sync your saved data;
• to authenticate you and keep accounts secure;
• to bill subscriptions and enforce plan allowances;
• to prevent, detect, and investigate abuse, fraud, and security incidents;
• to send transactional email about billing and your account — including email verification, password reset, and a security notice if repeated failed sign-in attempts pause your account — and, if you subscribed, product updates you can unsubscribe from at any time;
• to respond when you contact us;
• to find and fix faults in the app when something breaks, and to act on the feedback and bug reports you choose to send us;
• to understand aggregate usage (counts and trends, never content) so we can improve the product;
• to comply with legal obligations.

Where EEA/UK law applies, our legal bases are performance of a contract, our legitimate interests (security, abuse prevention, product improvement), your consent where we ask for it, and compliance with legal obligations.

• We do not sell your personal information, and we do not share it for cross-context behavioral advertising.
• We do not use your content to train AI models, and we do not grant our providers the right to do so on our behalf.
• We do not read your conversations. Our administrative tooling surfaces only metadata (counts, models, tokens, timestamps); it has no capability to return conversation content, and we keep it that way deliberately. The sole exception is a transcript you attached to a feedback or bug report yourself, which an administrator reads on the page where those reports are handled.
• We do not store your code or files. The product never persists them to our servers, and a transcript attached to a report carries brief previews of what the agent’s tools returned, not your files.

• AI model providers. To generate a response, the necessary conversation content (including any images or files you attach to a message) is sent to the provider you selected, such as Anthropic, OpenAI, Google, or xAI, under their API terms. We do not grant providers the right to train on this content.
• Service providers. Vendors that host and run the Services for us: Supabase (database and file storage), Stripe (payments), Google (Google sign-in, the Gmail API if you connect it, and the Gemini models when you select one), our email-delivery provider (transactional email), Vercel (website hosting and cookie-less site analytics), and Cloudflare (network delivery). They process data only to provide their service to us.
• Legal and safety. If required by valid legal process, or where reasonably necessary to protect the Services, our users, or others. Where lawful and practical, we will notify you.
• Business transfers. If Argenix is involved in a merger, acquisition, or asset sale, data may transfer with it; this policy continues to apply and we will notify you of material changes.

Argentic Code’s use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including its Limited Use requirements. Gmail access is used solely to send email you explicitly compose and direct. It is never used for advertising, is not read by humans, and is not transferred except as needed to provide that feature or comply with law.

• Account information: while your account exists.
• Conversations saved by the retired feature described above, including their attachments: until you delete them or your account, and no longer than it takes us to remove that content.
• Encrypted provider and connector credentials: until you disconnect them or delete your account.
• Usage counters, event records, and daily-activity history: for as long as reasonably necessary for security, abuse prevention, and aggregate analytics.
• Contact and support messages: while we handle them, together with any email replies we send you, which are stored with the message. Once a message is marked closed it is deleted about 6 months later (a review window), along with our stored replies. Newsletter emails are kept until you unsubscribe.
• Error diagnostics: each group of identical failures is deleted 90 days after the last time that failure occurred, so a fault that stops happening stops being stored.
• Feedback and bug reports, together with any session transcript attached to them: while your account exists. Deleting your account deletes them.

When you delete your account, your conversations, credentials, connectors, usage counters, error-diagnostic reports, and any feedback you sent (including the transcripts attached to it) are deleted, and everything that remains is irreversibly anonymized: the account record itself is reduced to an identifier-free tombstone, and audit entries and daily-activity history are severed from any identity so they no longer reference you.

Depending on where you live (including under the GDPR, UK GDPR, and California’s CCPA/CPRA), you may have the right to access a copy of your personal information, correct it, delete it, object to or restrict certain processing, and port it. We do not discriminate against you for exercising these rights, and we do not “sell” or “share” personal information as those terms are defined in the CCPA.

• You can delete your account yourself at any time — in the app under Settings → Account, or from your account page at argentic.argenix.ai. For anything else (or if you prefer email), write to support@argenix.ai from the address on your account; we will verify the request and respond within the time required by applicable law.
• Unsubscribe from the newsletter via the link in any issue or by emailing us.
• You can also lodge a complaint with your local supervisory authority or attorney general.

Automated decision-making. We do not make decisions about you by solely automated means that produce legal effects or similarly significantly affect you, and we do not profile you. The AI models in the product generate responses to what you ask them; they make no decisions about your account, your eligibility, or your rights. A few routine protections are automated, such as the temporary sign-in pause after repeated failed attempts and the enforcement of your plan’s monthly usage allowance, but these are limited and reversible, and you can reach a person about any of them at support@argenix.ai.

We encrypt data in transit (TLS) and encrypt stored provider and connector credentials at rest. We never store your password: only a one-way bcrypt hash of it, which cannot be reversed back into the password. Email-verification and password-reset links are stored only as one-way hashes too, are single-use, and expire. Our database enforces default-deny row-level security readable only by the backend service, and administrative access is restricted to a small allowlist.

Sign-in is rate limited and, after repeated failures on one account, paused for a short period, and we email the account owner when that happens. Resetting your password signs out every existing session on your account, and signing out revokes that session so it cannot be reused. You can turn on two-factor authentication, which asks for a code from your authenticator app in addition to your password. No system is perfectly secure; if a breach affects your data we will notify you as required by applicable law.

We operate from the United States. Your information is processed and stored in the US and in the locations of the AI model providers and service providers listed in section 7.

The Services are offered from and directed to the United States. We do not target the European Economic Area or the United Kingdom: we price only in US dollars, and we do not market, translate, or localize the Services for those regions. Where someone in those regions uses the Services on their own initiative, we still handle their information as described in this policy.

Where we transfer personal information out of the European Economic Area or the United Kingdom, we rely on the European Commission’s Standard Contractual Clauses, together with the UK International Data Transfer Addendum for transfers from the UK, incorporated into the data processing agreement we have with each provider. Where a provider is certified under the EU-US Data Privacy Framework, we may instead rely on that adequacy decision. You can request a copy of the clauses we rely on for a particular provider by writing to support@argenix.ai.

The Services are not directed to children under 13, and we do not knowingly collect their personal information. If you believe a child has provided us personal information, contact us and we will delete it.

When our data practices change — including when a new feature stores new kinds of data — we update this policy and its “last updated” date first, and for material changes we give reasonable notice (on the site, in the app, or by email).

Privacy questions and requests: support@argenix.ai.