Legal
Privacy Policy
Last updated November 11, 2026. Applies to everyone from that date.
Who we are
Gryz is run by GRTwo Digital LLC, a North Carolina limited liability company based in Chapel Hill, North Carolina, USA ("Gryz", "we", "us"). For personal accounts and our website, we decide how and why personal data is used, which makes us the controller. For company workspaces, see "Companies" below. You can reach us about privacy at info@grtwo.com.
What we collect
We try to collect only what we need to run Gryz. The main kinds of data are:
- Your content: the documents and Notes you write, publish or import, with their title, link, visibility and expiry; the memories, tasks, skills, projects and folders you or your connected AI tools save; and comments. To make search work, we also store text chunks and numerical embeddings made from memories and from documents and notes that are indexed.
- Email address: collected when you create an account (magic link, password, or Google/GitHub). Used for authentication and transactional email (welcome message, magic-link delivery, account notices). Where comments are available and you have not set a display name, the first part of your email address (the part before the @, capitalized) may be shown as your name on comments you post. You can set a display name in your account settings to choose what is shown instead. We never show your full email address or its domain.
- Network data: our hosting and security services see your IP address when you use the Service, and we use it for rate limits, bot checks and abuse prevention. We store a hash of it, not the address itself, for every publish, abuse reports, PIN attempts and error logs. A hash is hard to reverse but not impossible, so we treat it as personal data.
- Subscription data: if you buy a paid plan, Stripe processes your payment. We store your Stripe customer ID and subscription status (tier, renewal date, payment state). We do not store card numbers.
- Error data: when something breaks, we record the request path, a cleaned stack trace and, if you are signed in, your account ID. We design these reports to leave out document content.
- API keys: if you generate a key, we store the label you choose, a keyed hash of the key, and when it was last used. We do not store the key itself.
- OAuth MCP connections: if you authorize Claude or another MCP client, we store the registered client ID, granted scopes, and token expiry times. Access tokens are stored as one-way hashes only, never in plaintext.
- Connected AI tools: when you connect an AI tool, we receive what it sends through a tool call, and we store records of those calls (the tool used, which app called it, when and the outcome), not their arguments or responses. We do not read your local files or chat history.
- Connections to other services: if you connect an outside tool through Gryz, we store the access it gave you in encrypted form so your agents can use it.
- Terms acceptance: which version of our Terms you accepted and, for anonymous publishes, when.
- Other activity: likes, who you share with, setup progress, reports you file, moderation records, and feedback you send us.
- Companies, if you belong to one: your membership, role and project roles, invitations and the email address they went to, your last sign-in and whether two-factor sign-in is on, and the company's activity records. Activity records say who or which agent did what, with which tool, in which project, and the outcome. If the company turns on item records, they also say which items a connected app opened or listed, by id only. We also keep a one-day list of what each member read, to enforce read limits, and daily fingerprints (seals) of the records. If a legal hold keeps records after someone deletes their account, they show as an anonymous "Former member".
- Comments and notifications: where comments are available: the optional display name you set in account settings; the text of comments you post, with who posted them, when, and on which document; reports you file against a comment; the emoji reactions you leave on comments, which other people and agents see only as an emoji and a count, never who reacted, while your name and the emoji go to the person whose comment you reacted to, as a notification; and an in-app log of hearts, comments, and replies on your documents. If an AI agent connected to your account posts a comment, we also record that it came from an agent and whether it used an API key or an OAuth connection. That detail is kept internally and is not shown to other people.
We do not ask for your precise location, and we do not build profiles of your browsing on other websites. Our analytics providers receive some device and approximate location data, as described under "Third-party services". Please do not put sensitive personal data in Gryz unless you need to: public links can be opened by anyone who has them.
How we use it
Here is what we use data for and, for people in the EEA, UK and Switzerland, the legal basis for each use:
- To run the Service: storing and showing your content, search, sign-in, notifications, and sending you the emails the Service depends on. Legal basis: our contract with you.
- To keep the Service safe: rate limits, bot checks, abuse reports, automated and manual review of published content, and enforcing our Terms. Automated checks can block a publish or hold a document for review, and you can ask a person to look at the decision. Legal basis: our legitimate interest in a safe service, and legal duties.
- To manage subscriptions and accounting. Legal basis: our contract with you, and legal duties.
- To understand and improve the Service: analytics, session replay with text masked, and error tracking. Legal basis: your consent where the law requires it, otherwise our legitimate interest in improving the Service.
- To tell you about Gryz: setup tips and product updates by email. Legal basis: our legitimate interest, or your consent where the law requires it. Every one of these emails tells you how to opt out, and opting out does not stop emails the Service depends on, such as sign-in links and billing notices.
- To follow the law and protect rights: answering valid legal requests, reporting child sexual abuse material, and bringing or defending legal claims. Legal basis: legal duties and our legitimate interests.
AI and your content
We do not use your content to train AI models. Cloudflare, which computes search embeddings for us, states that it does not use customer content to train its models. When you connect an AI tool such as Claude, ChatGPT or Cursor, that tool's provider handles what the tool reads under your own agreement with that provider, including whether it trains on it.
Memories can be saved by your connected AI tools without you writing them yourself. You can see, change and forget them in Gryz.
Companies
A company workspace belongs to the company that runs it. For company content and company records, the company decides what is kept and why, and we handle that data for the company, on its instructions. For your account, billing, security, abuse checks and product analytics, we decide, and this policy applies as usual.
Company owners and admins can see company content, each member's name, role, sign-in email, last sign-in and two-factor state, connected apps, and the company's activity records. They cannot see your private space.
If you ask us about data in a company, for example to see it or delete it, we will send you to the company first and help the company answer. If the company asks, our support team may act on company content and settings. Those actions show in the company's records as "Gryz support".
Companies also get our Data Processing Addendum.
Third-party services
We use the following service providers to run the Service. Each gets the data it needs for its job and handles it under its own data protection terms. Stripe, Google and GitHub also act on their own account for payments and sign-in.
- Vercel (vercel.com): application hosting and edge infrastructure. All HTTP requests to Gryz (gryz.ai and, during the current brand transition, grtwo.app) pass through Vercel's network. Vercel may log request metadata (IP address, path, user-agent) for a short period for security and reliability purposes.
- Supabase (supabase.com): database, authentication, and session management. Stores document content, account data (email address), session tokens, and subscription records.
- Google (accounts.google.com): identity provider. If you choose Sign in with Google, Google authenticates you and shares your email address with us through Supabase Auth. We do not receive your Google password.
- GitHub (github.com): identity provider. If you choose Sign in with GitHub, GitHub authenticates you and shares your email address with us through Supabase Auth. We do not receive your GitHub password.
- Stripe (stripe.com): payment processing. Handles billing for Pro and Pro Plus subscriptions. We store only your Stripe customer ID and subscription status. No card numbers are stored by us.
- Resend (resend.com): email. Sends sign-in links, account and company notices, setup tips and product updates. Your email address and the message go to Resend for this.
- Cloudflare (cloudflare.com): AI inference for search (Workers AI). The text of the memories you save, and the queries used to search them, are sent to Cloudflare to compute numerical embeddings. When search over your documents and notes is turned on, the text of your documents and notes, and the search queries you or your connected AI tools make, are sent to Cloudflare too, to compute embeddings and rank results. Cloudflare states that it does not use customer content to train its models. The embeddings and text chunks are stored in our database (Supabase). You can turn document and note search indexing off in your settings and mark individual documents or projects as do-not-index.
- Upstash (upstash.com): rate limiting. Keeps short-lived counters so limits hold across our servers. A counter is named after an account, a key, or a network address, usually hashed. No document content.
- Cloudflare Turnstile (cloudflare.com): bot checks. When you publish without an account, your browser completes a Cloudflare challenge and we check the result with Cloudflare before saving the document. Cloudflare sees your browser and IP address for this check.
- Google Analytics (google.com): website traffic measurement on the production site. Receives the pages you visit (the page address and title, which for a shared document includes its short link and title), the page that referred you, your browser and device type, approximate location derived from your IP address, and the _ga cookie identifier. It does not receive your email address, your account ID, or document content. Google processes this data under its own terms.
- PostHog (posthog.com): product analytics and session replay. Receives page views, product events (for example, that a document was viewed, published or failed to publish, and its content type, size in KB, plan, and a reason code, that a shared-document call-to-action was shown or clicked, that an anonymous publish claim prompt was shown or dismissed, that anonymous documents were claimed after sign-up, that a document or collection was shared, which follow-up was used after publishing (such as copying the link or a prompt for your agent), which assistant was opened from the “Summarize with AI” links in the footer, that a subscription checkout was started, that an API key was generated, or that an account signed up or signed in), and screen recordings of how the Service is used, along with your account ID once you sign in. Events include an environment marker (production, preview, or local) and no document content. Page view events include the page address and title, which for a shared document include its short link and title. Session replay masks every text element and form field, so what you type, including document content, titles, and slugs, is never visible in a recording or sent to PostHog. Some events are sent from our servers rather than your browser: when an account signs up or signs in (with the method used, such as magic link, code, Google, GitHub, or password, and whether it started from a connector), when a publish through the website, API, or an AI client succeeds or fails (the surface used, your plan tier, content type, size in KB, and a reason code, with no account ID attached), and when an AI client calls a Gryz tool (the tool name, whether it succeeded, your plan tier, the self-reported name of the client, and how long it took). Tool inputs, results, and document content are never included. Email addresses, API key material, and document content are also never sent to PostHog. PostHog also gets error reports when something breaks: the error type, a code and where it happened. Reports from our servers carry no account ID. Reports from your browser are sent only if you allow analytics cookies.
AI apps you connect are not our sub-processors. When you connect an AI app such as Claude, ChatGPT or Cursor to Gryz, what it reads goes to that app's provider because you (or your company) chose to send it there, under your own agreement with that provider. In a company workspace, the company decides whether apps its members connect can reach company data, and Gryz keeps a record for the company of which app used which tool and, if the company turns it on, which items it opened or listed. Never the content.
If you have questions about a specific provider, contact us at info@grtwo.com.
Our sub-processors page lists these services and where they handle data. We update it before a new sub-processor starts, and we email company owners at least 30 days before a new one handles company data.
Where we store data
We are based in the United States, and we store your data there. We do not offer hosting anywhere else. Some providers run global networks, so a request can first be handled near you, and a provider may process data in another country. When you use Gryz from outside the United States, your data goes to the United States, where privacy laws may differ from yours.
For personal accounts, this transfer is needed to give you the Service you ask for. Where the law requires a safeguard for our providers or for companies, we rely on the European Commission's Standard Contractual Clauses and the UK and Swiss equivalents, in our providers' data protection terms and in our Data Processing Addendum for companies.
Data retention
We keep personal data for as long as we need it for the purposes in this policy, for as long as your account is open, and as the law requires. These are the specific periods:
- Documents: kept until you delete them or their expiry date passes. Deleted documents are removed from our live database right away. Their earlier versions go with them. If you are on a paid plan and it ends, your documents stay.
- Notes: deleted Notes are removed after 30 days.
- Accounts: retained until you request deletion. You may delete your account directly from your account settings page, or by emailing info@grtwo.com. Deletion is completed within 30 days of your request. Your documents and Notes in your own space, and any paid subscription, end with your account. If you are the only owner of a company, you need to add another owner before you can delete your account.
- Personal memories: retained until you, or an AI tool you connected, forget them, or your account is deleted. When a memory is forgotten, its text, its earlier versions, any passage it quoted from a document and any topic name only it used are removed from our live database right away, and it leaves the queue for Cloudflare. When a team member forgets a memory, an earlier version someone else wrote stays until you forget it. We keep a record that something was forgotten and when, without the text. Copies in database backups or held by our service providers are outside this and follow their own schedules. Company memories follow the company content line below, and cannot be forgotten under a company legal hold. A copy you brought into a company is company content: it stays until it is removed there.
- Company content: belongs to the company, not to the person who wrote it. It stays in the company when that person leaves or deletes their account, until the company deletes it or the company is deleted.
- Company records: lines about agents and connected apps are kept for 365 days. Other lines are kept for as long as the company exists. During a legal hold they are kept longer. Deleting an account removes the person's name from them; during a legal hold they show as an anonymous "Former member" instead. Deleting a company removes its content and records.
- Hashed IP addresses: kept for up to 90 days, then removed automatically. The one exception is a keyed hash we keep for as long as it is needed to enforce against repeat abuse by anonymous publishers.
- Comments: retained until you or the owner of the document delete them, or until the document is deleted or expires, or your account is deleted. A deleted comment is blanked immediately and its reactions are removed. A reaction you take back is removed straight away, along with its notification if unread. A notification already read is kept for 90 days.
- Notifications: read notifications are deleted after 90 days.
- Error logs: kept for up to 90 days, then removed automatically.
- Records of AI tool calls and of calls to connected tools: kept for 60 days. Daily totals of AI tool calls stay longer.
- Anonymous publisher tokens: expire after 7 days.
- Subscription records: we delete ours with your account. Stripe keeps its own payment records for accounting and tax purposes.
- Backups and provider copies: copies of data can stay in backups and in our providers' systems for a period after it is deleted from our live database. We do not use them except to recover from a failure, and they follow their own schedules.
Security
We use reasonable technical and organizational measures to protect your data, such as encrypted connections, access controls and storing keys and tokens as hashes. No system is perfectly secure, so we cannot guarantee that your data will never be accessed without permission. If a breach affects your personal data, we will tell you when the law requires it. You are responsible for keeping your sign-in details and API keys safe.
Your rights
Depending on where you live, you may have the right to access your data, get a copy of it, correct it, delete it, limit or object to how we use it, and withdraw consent. You can do much of this in your account. To exercise any of these rights, email us at info@grtwo.com. We will respond within 30 days, or sooner if the law requires. We may need to confirm who you are first, and we may keep data the law requires us to keep.
For data in a company workspace, the company decides, so please ask your company first. We will help it answer.
If you are in the EEA, UK or Switzerland, you can also complain to your local data protection authority.
If you live in a US state with a privacy law, you can appeal a decision we make about your request by replying to it, and an authorized agent can act for you where that law allows. We will not treat you differently for using these rights.
Children
Gryz is not directed at children under 13, and we do not knowingly collect personal data from them. Where the law sets a higher age for agreeing to an online service, you need a parent or guardian's agreement until you reach that age. If you believe a child has given us personal data, contact us at info@grtwo.com and we will delete it.
Changes to this policy
We may update this policy. For material changes, we will email the address on your account at least 30 days before they apply to you. Changes apply to new users when we publish them. The date at the top of this page shows when this version was last updated.
Contact
GRTwo Digital LLC, Chapel Hill, North Carolina, USA. Questions about this policy? Email us at info@grtwo.com. We aim to respond within 2 business days.