# Privacy Policy - Relex > Relex Privacy Policy # Relex Privacy Policy **Last updated:** 30 October 2026 **Effective date:** 30 October 2026 **Service:** Relex (`https://relex.legal`) **Support:** `support@relex.legal` **Privacy / legal contact:** `legal@relex.legal` --- ## 1\. Who We Are This Privacy Policy explains how Relex collects, uses, stores, shares, and otherwise processes personal data in connection with: - the website `https://relex.legal`; - any web or mobile applications we operate; - case-management, research, drafting, review, collaboration, and AI features; - the marketplace and provider directory; - the Relex MCP connector (`https://relex.legal/api/mcp`), which lets external AI assistants you choose to connect work with your Relex account (see Section 6.9); and - support, security, billing, onboarding, verification, and compliance activities. The controller for the personal data described in this Privacy Policy is: **Harmoniis Agents Ltd** Registered office: **Office 17108, 182-184 High Street North East Ham, London** Registration number: **16990810** Email: `legal@relex.legal` If a separate affiliate or local entity acts as controller for a specific service or country, we will identify that in the relevant service flow or notice. --- ## 2\. Roles: Relex, Lawyers and Clients Relex provides a software platform. It is not a law firm, does not provide legal services, and does not act for or on behalf of any lawyer, notary or other Provider who uses it. ### 2.1 Relex as controller Relex is the **controller** of the personal data it processes to run the platform, including the account and platform data it keeps for clients and for lawyers and other Providers. This includes personal data used to: - create and manage user accounts; - store and display matters, documents and messages on the platform, and run the platform's AI Features (Section 6); - provide subscriptions and billing; - run the public website and provider directory; - perform fraud prevention, security, moderation, and abuse handling; - verify providers and manage platform trust and safety; - respond to support, complaints, and legal requests; - send essential service communications; and - comply with legal, tax, accounting, and regulatory obligations. ### 2.2 Lawyers and other Providers are separate controllers A lawyer or other Provider who works with a client through Relex decides independently what to do with the client's matter and how to use the client's information in their professional work. They do so as a separate, independent controller, under their own professional and data-protection obligations. Relex does not take those decisions and is not the lawyer's processor. Each party is responsible for its own processing: Relex for the platform, and the lawyer for their professional work. If you are a client, questions about how a lawyer uses your information in their work should go to that lawyer; questions about the platform should go to Relex. --- ## 3\. Categories of Personal Data We Collect Depending on how you use the Services, we may collect the following categories of personal data. ### 3.1 Identification and account data - name; - email address; - phone number; - username; - organization name; - job title; - account credentials and authentication metadata. ### 3.2 Billing and transaction data - billing address; - tax identifiers where required; - subscription status; - payment-related metadata; - invoice data; - transaction references; - payout and connected-account data when you are a Provider. ### 3.3 Provider verification and trust-and-safety data - identity-verification status; - professional-status or credential-verification data; - business registration information; - licence, chamber, bar, or notarial registration information where applicable; - sanctions, fraud, or risk-screening results; - moderation records, complaint records, and review-related information. When you register as a Partner who must hold a regulated professional licence (lawyer or notary in jurisdictions where licence verification is mandatory), Relex's automated verification agent transmits your legal name and licence number to the public lookup tool maintained by the issuing regulator (for example, the Solicitors Regulation Authority in England & Wales, the State Bar of California, the Bundesrechtsanwaltskammer in Germany). The transmission targets only the regulator's official register; the resulting record — including the page URL we checked and a short excerpt — is retained in Relex's audit log for compliance, fraud-prevention and dispute-resolution purposes. Where the regulator does not publish an automatable register, the same information is reviewed manually by a Relex teammate. We do not share this data with third parties other than the regulator and our infrastructure sub-processors listed in this policy. ### 3.4 Usage, device, and log data - IP address; - browser and device information; - app and session identifiers; - language, timezone, and approximate geolocation derived from IP; - pages viewed, clicks, and feature usage; - error logs, diagnostics, and security logs. ### 3.5 Content you provide - prompts and inputs to AI Features; - uploaded files and documents; - messages and communications; - matter notes, tasks, contacts, and metadata; - support tickets and correspondence. ### 3.6 Special categories and sensitive data Because Relex is used in legal, regulatory, and compliance contexts, users may choose to upload or input data that is sensitive under applicable law, including health information, official identification numbers, criminal-offence data, or other special-category data. We do not require such data unless necessary for the intended service or legal compliance, and we ask users not to upload more personal data than needed. ### 3.7 Children The Services are intended for adults and are not directed to anyone under 18\. You must be at least 18 to create an account (Terms of Service, Section 5). We do not knowingly collect personal data from children as users of the Services. If you believe a child has created an account or given us their personal data, contact `legal@relex.legal` and we will delete it. Matter content uploaded by a user may concern a child, for example in a family-law matter. That data is processed only on the user's instructions for that matter, under Section 3.6 and Section 12. --- ## 4\. Sources of Personal Data We collect personal data: - directly from you; - from your organization or account administrator; - from users who invite you into a workspace or share data with you; - from Providers and Customers interacting through the marketplace; - from payment and verification providers, including Stripe; - from public registries, professional bodies, sanctions sources, or official verification sources; - from cookies and similar technologies; - from third-party integrations you choose to connect (see Section 16); and - from external AI assistants you connect over the Relex MCP connector, which send us the requests you or they make on your behalf (see Section 6.9). --- ## 5\. Purposes of Processing and Legal Bases Under the UK General Data Protection Regulation and Data Protection Act 2018 ("UK GDPR"), the EU General Data Protection Regulation ("EU GDPR") and similar laws, we process personal data only where we have a lawful basis. ### 5.1 To provide the Services We process data to register accounts, authenticate users, provide subscriptions, host workspaces, enable case-management and document features, provide AI Features, display provider profiles, and operate core functionality. **Legal basis:** performance of a contract; legitimate interests. ### 5.2 To process payments and subscriptions We process data to charge subscription fees, manage invoices, handle payouts, prevent payment fraud, and maintain billing records. **Legal basis:** performance of a contract; legal obligation; legitimate interests. ### 5.3 To verify Providers and protect marketplace integrity We process data to verify identity, review professional status, assess risk, prevent fraud, manage badges, respond to reports, and suspend or delist unsafe or misleading Providers. **Legal basis:** legitimate interests; legal obligation where applicable; substantial public interest or another applicable condition where sensitive verification data is involved and local law requires one. ### 5.4 To operate AI Features We process prompts, documents, and outputs to generate responses, summaries, extracts, draft text, and other requested functionality. **Legal basis:** performance of a contract; legitimate interests. ### 5.5 To provide support and communicate with you We process data to answer questions, resolve issues, send service notices, respond to legal requests, and communicate about your account. **Legal basis:** performance of a contract; legitimate interests; legal obligation. ### 5.6 To secure the Services We process data for logging, incident response, authentication, abuse detection, fraud prevention, monitoring, backup, rate limiting, and vulnerability management. **Legal basis:** legitimate interests; legal obligation where applicable. ### 5.7 To improve and develop the Services We use analytics, telemetry, diagnostics, and limited service-improvement data to understand usage patterns, improve reliability, and develop features. **Legal basis:** legitimate interests; consent where required by law. ### 5.8 To comply with legal obligations We may process data for accounting, tax, sanctions compliance, law-enforcement response, court orders, consumer-law compliance, and regulatory reporting. **Legal basis:** legal obligation. ### 5.9 Marketing We may send product updates, offers, newsletters, or event invitations where permitted by law. When you create an account, your name and email address are added to our email provider's (Resend) customer list so that we can send you service and product emails. **Legal basis:** consent where required; legitimate interests where permitted for existing business contacts. You can opt out at any time. --- ## 6\. AI and Automated Processing Notice ### 6.1 AI-assisted features Some parts of the Services use AI-assisted technologies to generate, review, summarize, extract, classify, compare, or suggest content. ### 6.2 No solely automated legal decision-making by Relex Relex does not intentionally use solely automated decision-making that produces legal or similarly significant effects on users unless expressly disclosed and lawfully implemented. ### 6.3 Human review and quality control We may use human review, testing, abuse monitoring, and audit logging to improve reliability, security, and compliance. ### 6.4 No model training on customer matter content We do not use customer matter content, provider-client matter content, prompts, uploaded documents, or anything a connected AI assistant sends us (Section 6.9) to train, fine-tune, or improve AI models, whether ours or any third party's, and we do not permit our AI subprocessors to use it to train theirs. We will not change this default without first providing you express, prominent, prospective notice and, where required by law, obtaining your prior consent. Service-improvement work permitted under Section 5.7 uses analytics, telemetry, diagnostics, and aggregated usage signals — not matter content. ### 6.5 Use of third-party AI subprocessors AI Features are powered by third-party AI subprocessors (large-language-model, speech-to-text and embedding providers) selected by Relex. Subprocessors process data submitted to AI Features (including prompts, inputs, attached documents, and metadata) to return outputs to the User. We have not given any AI subprocessor permission to use this data to train its models. Under Google Cloud's terms, Google does not use data sent to Vertex AI to train or fine-tune its models without the customer's permission. The provider composition may change; we will update this list when it does. Current AI subprocessors: - **Google Cloud (Vertex AI)** — runs the legacy server-side document reader used for documents read on Relex servers before browser reading became the default (see Section 6.6), the Google Gemini models used by the Relex AI agents and AI Features, and document retrieval (search over your matter documents and Relex's legal corpus). - **OpenRouter, Inc.** — routes requests to AI models. Every model request Relex sends through OpenRouter is restricted to model providers that OpenRouter lists as zero-data-retention, with OpenRouter's data collection setting set to deny, so it is not sent to a provider that may store or train on it. If no such provider is available for a model, the request fails rather than going to another provider. This covers the Relex AI agents and AI Features, page reading of uploaded documents, audio transcription (by a Google Gemini model), reading of images and audio inside uploaded files, and the decision model that steers the agents. Under this restriction, the Google Gemini models used by the agents, AI Features and audio transcription are currently served only by Google Cloud Vertex AI. - **Web search for legal sources.** When an agent searches the web or opens a web page to find legal sources, OpenRouter passes the search query or web address that the model writes to its web search and fetch service, currently provided by **Firecrawl**. OpenRouter's zero-data-retention restriction does not apply to that service; the query or address is handled under Firecrawl's own terms. Because the agents receive only the redacted, placeholder form of uploaded documents (Section 6.6), these queries are not written from raw document content. Do not include personal data in a request you ask an agent to research on the web. If your organisation configures its own redaction model endpoint in Relex settings, document content sent for redaction goes to that endpoint instead. That endpoint is chosen and controlled by your organisation, not by Relex. External AI assistants that you connect to Relex yourself (see Section 6.9) are not Relex subprocessors: they act for you, under your own agreement with their provider. Organisational customers may request further detail by contacting `legal@relex.legal`. ### 6.6 Architectural separation of personal data from AI Features We separate personal data from the AI Features by design: - **Structured surfaces (no AI exposure).** The Parties registry and the in-chat party dialogs write directly to the Relex database. The AI Features never receive their content. Real names, identifiers, contact details, addresses, and other personal data entered through these surfaces are held in your encrypted private-information vault (see Section 12.2), decryptable only in your browser with your PII password and visible only to you and to other principals you authorise. - **Drafts at rest.** Generated drafts (claims, complaints, contracts, notices, etc.) are stored on Relex infrastructure with placeholder tokens — for example, `[PARTY_NAME_1]`, `[PARTY_ADDRESS_1]`, `[NATIONAL_ID_PLACEHOLDER_1]` — and never with real names or identifiers. The AI Features see only the placeholder form of any draft they generate or modify. - **Substitution happens in your browser.** When you open, copy, email, or download a draft, the substitution of real values for placeholders is performed on the fly by your browser, in memory, using party records fetched from the Relex database. No substituted form is persisted on the server. - **Chat-input guardrail.** A client-side detector intercepts the most common categories of personal data (national identifiers, tax identifiers, IBANs, telephone numbers, and email addresses) when they appear in the free-text chat input, redirecting you to a structured party form before any message is sent. A server-side detector runs the same checks as a fallback. Together these prevent the most common accidental disclosures. - **Uploaded document content.** Document content is read and redacted in your browser, on your device, before it reaches any server. Your browser downloads Relex's reading models from Hugging Face (a fast text reader and, on devices that support WebGPU, a higher-accuracy reading model, installed in the background after you first sign in) and runs them locally. Pages these models cannot read reliably are kept in redacted form and flagged for your review, or left out; they are not sent to a server to be read. If you choose to add your own AI model endpoint in your settings, pages that need it are sent by your browser directly to the endpoint you configured, under your own arrangements with its operator, and not through Relex servers. Documents uploaded before browser reading became the default may have been read on Relex servers by a reading model running in Relex's Google Cloud project (Vertex AI) or routed through OpenRouter to zero-data-retention providers only; those models saw the raw content only to read and redact it, under Relex's data-processing terms (see Section 6.5). Formats your browser cannot read yet, such as audio and some office or legacy formats, are kept only as an encrypted original and are not read or processed. In every mode, only the redacted, placeholder form is made available to general-purpose AI Features and to any connected external AI assistant (including Claude, ChatGPT, Grok, Gemini or Perplexity, when connected over our MCP connector). Those downstream features and assistants never receive raw document content directly; you can also always choose to view a document yourself in your browser via a secure link instead. Redaction is automated and can miss identifiers. The chat-input guardrail is a best-effort safety net and not a guarantee. **Where personal data does reach the AI Features because you typed it into the chat input (whether by ignoring a warning, bypassing the detector, or because the detector did not recognise the pattern), that data is processed by the AI subprocessors under their respective terms, and you bear sole responsibility for that disclosure** — see Section 6.7 of the Terms of Service. ### 6.7 Relex AI agents Relex runs its own AI agents on Relex servers: - an **evaluation agent**, which reviews a new matter, asks follow-up questions and produces an evaluation, a service tier and, for clients invited by a lawyer, an engagement offer; and - a **case agent**, which works on an active matter when you, or an external AI assistant you have connected, ask it to (for example research, drafting or re-assessing the matter), and saves its results to the matter. These agents work on the placeholder form of the matter and never on the sealed party identities held in your private-information vault. They use the models listed in Section 6.5\. Their outputs are drafts, not legal advice. A lawyer on the matter can validate a draft, and that validation is recorded in Relex, but drafts are available on the matter whether or not they have been validated. Relex does not guarantee that a lawyer reviews every output unless your engagement with that lawyer says so. ### 6.8 Clients invited by a lawyer (guest intake) If a lawyer using Relex invites you as a client, you describe your matter to the Relex AI agent; the guest pages tell you that you are talking to an AI. The agent produces the evaluation, service tier and engagement offer. You then receive the lawyer's engagement agreement to sign electronically and pay the lawyer's own fee through the lawyer's payment account. Relex provides the platform and the AI agent, and stores your account and matter data on the platform as controller (Section 2.1). Your matter content is processed by the Relex AI agents described in Section 6.7\. Relex does not provide legal services and does not act for the lawyer. The lawyer decides independently how to handle your matter and is responsible, as a separate controller, for their professional work and for how they use your information in it (Section 2.2). The engagement agreement is between you and the lawyer. ### 6.9 External AI assistants connected over MCP (connectors) You can connect an external AI assistant — for example Claude, ChatGPT, Grok, Gemini or Perplexity, or another client that supports the Model Context Protocol (MCP) — to your Relex account at `https://relex.legal/api/mcp`. You do this yourself: the assistant sends you to a Relex sign-in and consent page, and the connection is created only after you sign in and approve the permissions (scopes) shown. **Scopes.** The consent page lists each scope the assistant asks for: - `relex.cases.read` — read your matters, ontology, timeline and drafts (structure and placeholder form only, never your clients' personal data); - `relex.cases.write` — read, create, update and delete permitted resources, and link parties to a matter by reference; - `relex.draft` — draft and revise legal documents; - `relex.agent` (optional, never requested by default) — broad access: read, create, modify and delete permitted workspace records and delegate work to the Relex AI agents. With this scope your assistant may not ask you to confirm each write action. Assistants request the first three scopes by default. Grant only what you need. **What a connected assistant can read.** Matter references and links, the redacted matter context (parties appear only as labels such as `[PARTY_NAME_1]`), source readiness, paged redacted document text, earlier drafts and session conclusions, and published, verified marketplace profiles. Requests that would return your clients' real identities from your private-information vault, or an original unredacted document, are refused and answered with a link to open the item yourself in the Relex web app. **What a connected assistant can write.** Within the scopes you granted, it can save drafts and new revisions, record evidence-backed corrections to a matter's ontology, save the conclusions of a session it opened, ask the Relex AI agents to work on a matter (Section 6.7), prepare a request to a legal professional (Relex returns an approval link; nothing is sent, shared or charged until you approve it in Relex), and create or update records through documented Relex API endpoints. Deletions and other irreversible actions, such as marking an invoice paid, are not carried out over the connector; Relex returns a link where you confirm them in the web app. The one exception is removing a party's link from a matter. **What Relex receives from the assistant.** Relex receives only the requests the assistant sends to the connector: the tool it calls and the inputs for that call, such as a matter reference, a search query or the text of a draft to save. Relex does not receive or ask for your conversation history, the assistant's memory, your other chats or files in the assistant, or your account details with its provider. We use what we receive only to carry out the request. We do not use it to build a profile of you, for advertising, or to train AI models. Content you ask the assistant to save (for example a draft or session conclusions) is stored as matter content and kept under Section 11\. Our logs of connector calls record the tool name, the type of client, the outcome and timing, not the content of requests or responses. **Where the data goes.** Whatever the assistant reads from Relex is sent to that assistant's provider and is handled under your agreement with that provider and its privacy policy. That provider is a separate, independent controller; it is not a Relex processor or subprocessor. Relex does not control how the provider stores or uses that data. The connector never returns your private-information vault (real names, national and tax identification numbers, contact details) or original unredacted documents, so those do not reach the assistant through Relex. **Tokens, logs and revocation.** Each connection is a separate Relex API key. Access tokens last 7 days and rotate; refresh tokens last 30 days; authorisation codes last 60 seconds. We keep a record of each connection (the client and the scopes granted) and process request metadata for security and rate limiting. You can revoke a connection at any time in **Settings › API keys**; it stops working immediately. --- ## 7\. Verification, Identity Checks, and Professional Status Checks If you are a Provider or apply to become one, we may collect and process personal data needed to verify your identity, business status, tax status, banking status, sanctions exposure, and professional or credential status. This may include checks performed by Stripe or other vendors, as well as checks against official registries or professional bodies. We may store the outcome of such checks, including statuses such as verified, pending, failed, expired, or suspended. --- ## 8\. Cookies and Similar Technologies We use cookies and similar technologies for: - essential site operation; - authentication and session management; - security and fraud prevention; - analytics and performance measurement; and - preferences and campaign measurement. **Essential storage.** This includes the `relex_session` sign-in cookie, the browser storage used by Firebase Authentication, a record of your cookie-banner choice, and a random identifier for the current browser tab (session storage). We use that tab identifier to link a tab's requests in our server logs when we investigate errors. **Relex usage analytics (consent-based).** Page views, time on page, scroll depth, clicks, page-performance measurements and feature use are sent only to Relex's own servers, and only after you accept analytics in the cookie banner. Once you accept, a random anonymous identifier is kept in your browser's local storage. If you arrive through a link in a Relex email or a link with campaign tags, the link's tracking identifier and campaign tags are kept in session storage for that visit so we can measure the email or campaign. **Error and reliability reports (no consent needed).** When a page error or a failed request occurs, your browser sends a short technical report to Relex's own servers: the error type, a shortened error message, the page address and the browser family. On dashboard pages, it also reports browser memory use. These reports are needed to keep the Services working and secure. They do not include an analytics identifier. **No third-party analytics, advertising or tag managers.** We do not use any third-party analytics service, including Google Analytics and Cloudflare Web Analytics, and we do not use Google Tag Manager, advertising pixels or other third-party marketing trackers. Our site's security policy blocks the Cloudflare Web Analytics script, so it cannot run in your browser even if our hosting provider adds it to a page. **Server-side monitoring.** Our servers log requests and errors to Google Cloud Logging and Google Cloud Monitoring (Section 9.1). Logs are kept for 30 days. Where required by law, non-essential cookies are used only with your consent. You can manage cookie choices through our cookie banner or browser settings. A separate Cookie Notice may provide more detail. --- ## 9\. Sharing of Personal Data We may share personal data with: ### 9.1 Service providers and subprocessors Hosting providers, cloud infrastructure providers, analytics providers, email and support providers, AI providers, payment providers, verification vendors, security vendors, and similar service providers that process data on our behalf. Our main providers are: - **Google Cloud** — application servers (Cloud Run, region europe-west1, Belgium), database (Firestore, United States multi-region), file storage (EU multi-region), backups, server logging and monitoring (Cloud Logging and Cloud Monitoring), and Vertex AI (Section 6.5); - **Google Firebase Authentication** — sign-in with Google, Apple or an emailed sign-in link; - **Cloudflare** — website hosting, content delivery and network security for `relex.legal`; - **Stripe** — subscription payments, invoices, and identity and payout checks for Providers (Stripe Connect); - **OpenRouter** — AI model routing (Section 6.5), including web search for legal sources through Firecrawl (Section 6.5); - **Resend** — service and product emails. When you use the address search in a party or matter form, the text you type is sent from your browser to the Photon geocoding service (`photon.komoot.io`, operated by komoot) to suggest addresses. Because documents are read in your browser (Section 6.6), your browser downloads the reading models from Hugging Face. ### 9.2 Other users and workspace participants Where the nature of the Service requires it, we may share your profile, communications, documents, or collaboration data with other users in your workspace or transaction flow. ### 9.3 Providers and Customers If you use the marketplace, relevant information may be shared between Customers and Providers to facilitate contact, service delivery, payment, verification, and dispute handling. ### 9.4 Public or semi-public profile information Provider profiles, badges, business descriptions, languages, jurisdictions, ratings, and similar directory information may be visible to other users or the public, depending on your settings and platform design. ### 9.5 Authorities and legal recipients We may disclose personal data to courts, regulators, law-enforcement bodies, tax authorities, professional bodies, or other third parties where required by law or reasonably necessary to establish, exercise, or defend legal claims. ### 9.6 Corporate transactions We may share or transfer personal data as part of a merger, acquisition, financing, restructuring, sale of assets, or similar transaction, subject to applicable confidentiality and legal protections. We do not sell personal data. We do not share it for advertising, including cross-context behavioural advertising, and we do not let third parties use it for their own purposes, except where the law requires us to disclose it (Section 9.5) or where you direct us to share it, for example with a legal professional or a connected AI assistant. --- ## 10\. International Transfers We may process personal data in countries outside the European Economic Area, United Kingdom, or other country where you are located. In particular, our application servers run in Google Cloud's europe-west1 region (Belgium) and uploaded files are stored in Google Cloud's EU multi-region, but the Relex database is stored in Google Cloud's United States multi-region, Cloudflare serves the website from its global network, and AI requests to Google Cloud Vertex AI and to providers reached through OpenRouter may be served outside the European Economic Area. Where we transfer personal data internationally, we use an appropriate transfer mechanism where required, such as: - an adequacy decision; - Standard Contractual Clauses; - the UK International Data Transfer Addendum or equivalent; or - another lawful safeguard. You may request more information about applicable transfer safeguards by contacting `legal@relex.legal`. --- ## 11\. Data Retention We keep personal data only for as long as necessary for the purposes described in this Privacy Policy, including to provide the Services, comply with legal obligations, resolve disputes, enforce agreements, and protect the Services. Retention periods depend on the category of data and purpose. For example: - account and billing records: retained for the period required by tax, accounting, and audit laws; - support records: retained for as long as reasonably necessary to resolve the issue and maintain support history; - verification and compliance records: retained for as long as necessary to manage trust, safety, fraud prevention, and legal obligations; - workspace and matter data: retained according to your plan, settings, contractual commitments, and legal requirements; - security logs: retained according to operational and security needs. Some periods are fixed by how the Services work: - **Account deletion:** when you delete your account in **Settings › Data**, deletion is scheduled for 30 days later. You can cancel during that time; after it, your account and its data are permanently deleted, except the evidence records described in Section 16A and records we must keep by law. - **Deleted files:** files you delete stay recoverable in storage for up to 90 days, then are purged. - **Raw uploaded images:** where a document is read on Relex servers (see Section 6.6), images uploaded for reading are kept in a temporary staging area only until they have been read and redacted, and for no more than 7 days. - **Backups:** daily database backups are kept for 7 days and weekly backups for 14 weeks. Encrypted disaster-recovery copies are kept in a separate, access-restricted backup project for 90 days and then deleted automatically. Once deleted, they stay recoverable for up to 90 more days and are then permanently purged, so a disaster-recovery copy is gone within about 180 days of being made. Deleted data can remain in backups until those backups expire. - **Server logs:** request and error logs are kept for 30 days. - **Connector tokens:** access tokens expire after 7 days and refresh tokens after 30 days (Section 6.9). - **Invitations:** workspace and partner invitations expire after 14 days. When the original purpose for processing has been fulfilled and no overriding legal, contractual, or legitimate interest requires further retention, we delete or anonymize the relevant personal data. --- ## 12\. Security ### 12.1 Technical and organizational measures We apply technical and organizational measures designed to protect personal data, including access controls, role-based permissions, logging, encryption in transit (TLS), encryption at rest, backup measures, provider due diligence, and security monitoring. No method of transmission or storage is completely secure, and we cannot guarantee absolute security. If a personal data breach occurs, we will notify the Information Commissioner's Office or other competent supervisory authority, and affected individuals, where the law requires it. Organisational customers will be told without undue delay where their data is affected. ### 12.2 How your data is stored — three different levels of protection Not all data in Relex is protected in the same way. It is important that you understand the difference: - **Your private-information vault — encrypted so that even Relex cannot read it.** The real identifying details of the people and organizations in your matters — names, national and tax identification numbers, and contact details — are held in a private-information vault that is encrypted **on your device**, in your browser, with a key derived from a PII password that only you know. Relex's servers store only the resulting ciphertext and non-identifying metadata. **Relex has no access to your PII password and cannot decrypt this vault.** This is a zero-knowledge design: the protection is cryptographic, not merely a policy promise. - **Everything else you work on — readable by Relex in order to run the Service.** Your documents, drafts, matter notes, tasks, and other working content are **de-identified** (real identifiers are replaced with placeholder labels such as `[PARTY_NAME_1]`) but are **not** encrypted under your PII password. This content is stored in readable form on Relex's infrastructure so that the Service — search, drafting, review, collaboration, and AI Features — can operate on it. It is protected by access controls, by our infrastructure provider's encryption at rest under keys that **Relex manages**, and by the other measures in Section 12.1\. In other words, Relex is technically able to read this content to provide the Service; you should not treat de-identified documents and drafts as hidden from Relex. - **A limited set of server-side secrets — encrypted under keys Relex manages.** Certain records that must be provable or protected independently — for example e-signature and notarial evidence, and the access tokens for third-party accounts you connect — are encrypted at rest under keys that Relex manages and can use when needed to operate the Service. ### 12.3 Your PII password and recovery key — losing both means permanent loss Because your private-information vault can be decrypted only with your PII password, **you are responsible for keeping that password — and the one-time recovery key we show you when you first set it up — safe.** - If you forget your PII password but still have your recovery key, you can restore access and set a new password. Your data is preserved. - **If you lose both your PII password and your recovery key, the contents of your private-information vault cannot be recovered by anyone — including Relex.** We hold no copy of your password or recovery key and have no technical means to decrypt the vault for you. The only remaining option is to reset the vault, which **permanently and irreversibly erases** its encrypted contents so that you can start again with a new password. - For an **organization** vault, access is shared by sealing it to each authorized member. Access survives as long as any one authorized member can still unlock it, or the organization passphrase is known. It is permanently lost only if every authorized member's access **and** the organization passphrase are all lost. This design is deliberate: encrypting your private-information vault under a key only you hold is what lets us guarantee that Relex — and any AI, including Claude — never sees your clients' real identities. The trade-off is that recovery of that vault depends entirely on you keeping your password and recovery key. If you believe your account or data has been compromised, contact `support@relex.legal` immediately. --- ## 13\. Your Rights Subject to applicable law, you may have the right to: - access your personal data; - correct inaccurate personal data; - delete personal data; - restrict processing; - object to certain processing; - receive your data in portable form; - withdraw consent where processing is based on consent; and - lodge a complaint with a competent supervisory authority. These rights apply under the UK GDPR and the EU GDPR. You can download a copy of your data and delete your account yourself in **Settings › Data**. You may also exercise any right by contacting `legal@relex.legal`. We respond within one month, which the law allows us to extend in some cases. We may ask you to verify your identity before responding. If you are in the United Kingdom, you can complain to the Information Commissioner's Office (`https://ico.org.uk`). If you are in the European Economic Area, you can complain to the data protection authority in your country of residence or work, or where you believe an infringement took place. We would appreciate the chance to address your concern first at `legal@relex.legal`. --- ## 13A. United States State Privacy Rights California, Colorado, Connecticut, Virginia, Utah, Texas, Oregon, Montana, and certain other US-state residents have rights under the California Consumer Privacy Act / California Privacy Rights Act (CCPA/CPRA) and analogous state privacy statutes, including: - the right to know what personal information we collect, use, disclose, and (where applicable) "sell" or "share"; - the right to delete personal information; - the right to correct inaccurate personal information; - the right to opt out of "sale" or "sharing" of personal information; - the right to opt out of targeted advertising; - the right to limit the use of sensitive personal information; - the right to non-discrimination for exercising these rights; and - where applicable, the right to appeal a denial. **No sale of personal information.** Relex does not "sell" personal information in the ordinary meaning of that term. We "share" data with service providers and AI subprocessors only to operate the Services on our behalf under contractual restrictions. **Sensitive personal information.** We use sensitive personal information only for the purposes described in Section 5 and do not use it to infer characteristics about you for advertising. **Exercising rights.** Email `legal@relex.legal`. We may verify your identity using account information. Authorized agents may submit requests with appropriate verification. **Notice of financial incentive.** We do not offer financial incentives in exchange for personal information. --- ## 14\. Provider and Customer Responsibilities for Data They Upload Users are responsible for ensuring they have a valid legal basis for uploading, sharing, or otherwise using personal data through the Services. If you are a Provider, you remain responsible for your own professional, confidentiality, and data-protection duties to your clients unless a separate written agreement states otherwise. Lawyers, other Providers and organisational customers are separate controllers for their own use of personal data in their work (Section 2.2). Each is responsible for its own legal basis, transparency to its own clients and other data subjects, and any required impact assessments. --- ## 15\. National Identification Numbers and Similar Sensitive Identifiers We collect government-issued identification numbers, such as passport numbers, national ID numbers, or equivalent national identification numbers, only where necessary, proportionate, and legally justified for onboarding, identity verification, compliance, payment, or similar lawful purposes. Please do not provide such identifiers in free-text fields or general support tickets unless specifically requested through a secure flow. --- ## 16\. Third-Party Websites and Integrations The Services may link to third-party websites or services or allow you to enable third-party integrations. In **Settings › Connectors** you can connect your own accounts — Google Drive, a practice mailbox (IMAP/SMTP), Twilio (voice and SMS) or WhatsApp Business — so that Relex can bring documents and messages into your matters. Relex stores the access token or credentials for each connection encrypted (see Section 12.2) and uses them only for the connection you set up. You can disconnect at any time. External AI assistants connected over MCP are described in Section 6.9. We are not responsible for third-party privacy practices. Review the privacy notices of those third parties before using them. --- ## 16A. Evidence, Signatures, and Legal Retention Relex creates evidence records for events that may need to be proved later, including account creation, proof of email ownership, sign-in provider evidence such as Google or Apple authentication, partner email verification, Stripe subscription and Connect references, merchant payment references, case activation consent, partner advisor input, and partner document validations. These records may include personal data such as IP address, user agent, timestamps, account identifiers, email-verification status, payment-provider references, consent text or consent hashes, document hashes, document storage versions, and partner identity or professional-verification status. We keep these records separate from ordinary profile content and restrict their use to security, compliance, payment disputes, arbitration, legal claims, regulatory matters, and court evidence. Where applicable law allows or requires retention for legal claims or compliance, these evidence records may be retained after deletion of other account data. --- ## 17\. Changes to This Privacy Policy We may update this Privacy Policy from time to time to reflect changes in the Services, law, our processors, or our practices. We will post each updated version on this page and change the "Last updated" and "Effective date" lines at the top. If we make material changes, we will also tell registered users by email or in-app notice before the changes take effect. We will not make a material change to how we use data you have already given us, such as using matter content for AI training, without first giving you that notice and, where the law requires it, asking for your consent. --- ## 17A. Language; controlling version This Privacy Policy is drafted in the English language. Any translation of this Privacy Policy is published for the convenience of data subjects only. In the event of any discrepancy or conflict between the English-language version and any translation, **the English-language version shall prevail**, save to the extent that mandatory consumer-protection law in the data subject's country of residence requires the local-language version to govern the consumer relationship. --- ## 18\. Contact Us For privacy, data protection, or legal questions, contact: - `legal@relex.legal` - `support@relex.legal` **Mailing address:** **Office 17108, 182-184 High Street North East Ham, London** If you are in the UK, you may also contact the Information Commissioner's Office. If you are in the EEA or Switzerland, you may contact your local data protection authority. --- _[View this page on relex.legal](https://relex.legal/privacy-policy)_