Senova Privacy Notice

When you chat or call the API, your message, the recent conversation and any attached files are sent to the model provider serving that model (see subprocessors) to generate the reply. Each provider processes this data under its own API terms to produce the reply. In Senova Code, the model provider also receives the project files, command output and package type information (read from the packages installed in your project) that Senova Code uses in that conversation. Staff can see usage totals, and see chat content only when you chose to help improve Senova (below), when you share a chat with us for support, or when the law requires.

Senova can remember useful facts about you, such as your name, your preferences, people you mention and your work, so you don't have to repeat them. Learn from my chats is on by default. After the first message in a chat, then every few messages and when you leave, Senova reads what you wrote and saves lasting facts, shows a “Saved to memory” note you can undo, and uses them in later chats. You can also ask it to remember or forget something, or add facts yourself. It never saves passwords, keys, card or ID numbers, and it saves health details or beliefs only if you add them yourself.

See, change or delete it. Settings → Memory lists everything Senova remembers. You can edit, pin or delete any item, export all of it (JSON or Markdown), or Forget everything. Turn it off with Learn from my chats in Settings → Memory; nothing new is learned after that, and what's already saved stays until you delete it. Chats you mark “Don't learn” are skipped.

Memory is sent to the model provider with your messages as context for the reply, and may be sent to our embedding provider to pick the most relevant items (see subprocessors). It is separate from Improving Senova: turning memory on doesn't let us use your chats to improve Senova, and staff don't read your memory except as described in AI processing above.

Senova gets better by learning from real conversations. If you choose to help, we may use the chats you start from then on to find mistakes, measure the quality of answers, and improve Senova's prompts, features and the models it runs. This can include Senova staff reading a chat, and running it through our own testing tools. We never sell your chats or use them for ads.

Failed tasks. When a task in one of these chats fails (a step breaks, Senova runs out of steps, you press Stop or give a reply a thumbs down), we keep a test case: your request with your own name, contact details, card and ID numbers and secrets masked and links cut back to the website name, the names of the tools Senova used, and how it failed, with the day but not the time. Only Senova staff can read it; they may add it to our test set or discard it. Unreviewed cases are deleted after 90 days, and deleting the chat (or your account) deletes its case. We also count these tasks, without any content, for our public task scoreboard.

Training examples. When you give a reply in one of these chats a thumbs up or down, we may keep your request and Senova's reply as an example for training the models Senova runs, masked the same way (your own name, contact details, card and ID numbers, secrets and links), with the day but not the time and nothing that says who you are. We don't keep chats that look sensitive or replies that used tools (such as email, calendar or the browser). Senova staff read each example and only the ones they approve are used for training. Unreviewed examples are deleted after 90 days, rejected ones after 30 days and approved ones after 18 months. Deleting the chat or your account deletes its examples, and turning Help improve Senova off deletes every example from your chats straight away. A model we already trained can't forget what it learned, but your examples are left out of every training run after that.

Your choice. Everyone chooses before anything is used: new accounts on the sign-up form, and any account that hasn't chosen yet (such as accounts created before 10 November 2026) on a one-time question when it next signs in. In the United States the choice starts as yes; in the EU, the EEA, the UK, Switzerland, everywhere else, and wherever we can't tell where you are, it starts as no and you have to turn it on. Until you choose, nothing is used. You can change it at any time in Settings → Privacy & data; a change applies to chats from then on. We keep a record of your choice (see the table above).

Never used: chats from before you said yes, temporary chats, chats in team or business workspaces, calls to the developer API, conversations with agents you publish for your own customers, chats our safety checks flag as containing sensitive information (such as health, financial or identity details), and anything you delete. Deleting a chat ends its use. We also never use replies written by Anthropic's Claude or OpenAI's models to train models.

Please don't put anything in a chat you wouldn't want reviewed, such as health details or other people's personal information. Where the law requires a legal basis (for example in the EU or UK), ours is your consent, which you give by saying yes and withdraw by turning it off.

When you rate a reply, report a bug or send a suggestion, we keep what you write (and any reasons you pick), up to three screenshots if you add them, and some details about where you were: the page address without anything after it, the part of the app, your browser's name and version, the app version and, for a reply, the Senova model name. We re-save screenshots when they arrive, which removes their location and camera details, so crop out anything private first.

The conversation is shared only if you tick “Include this conversation”, which is off unless you turn it on. Then our staff can read the recent messages of that chat (up to the reply you rated, if you rated one) for as long as the chat exists; we don't make a copy, so deleting the chat ends their access. Each time staff open a shared conversation, it's recorded.

We use your report to find and fix problems, decide what to improve and reply to you. Only Senova staff see it, through accounts protected by two-factor sign-in; it isn't sent to an AI model or anyone else. When we reply, you see our message in Settings → My reports, and we may also email it to you. The email has our reply and a link, not your report.

You can delete any of your reports, with its screenshots, at any time in Settings → My reports, or email hello@senovasys.com and we'll delete it for you. Otherwise we keep a report while we work on it and delete it, with our replies, 18 months after we close it. Screenshots are deleted 90 days after the report is closed. Deleting your account deletes all of your reports.

We use only strictly necessary cookies: your sign-in session and, briefly, cookies that protect sign-in with Google. No analytics, advertising or tracking cookies.

Only the subprocessors that run the Services, members of your organization as your role allows (they never see your chats), and authorities where the law requires. We don't sell or share personal data for cross-context behavioral advertising.

Download your data from Account → Your data, delete chats at any time, turn the use of your chats to improve Senova on or off in Settings → Privacy & data, and delete your account in Account settings. Depending on where you live (for example California and other US states, the EU or the UK) you may also have rights to correct data, object to or restrict processing, and complain to a regulator. Email hello@senovasys.com and we'll answer within 30 days (45 days at most where US state law allows).

Data is encrypted in transit; access is limited by role and enforced in the database; passwords and API keys are stored as one-way hashes. The Services are for people 18 and over. Data is stored in the United States. We'll tell you about material changes to this notice before they take effect.