Privacy Policy

LiveContext is committed to protecting your privacy. This policy explains how we collect, use and safeguard your personal information when you use our platform.

By using LiveContext, you agree to the practices described here. If you have questions, contact us at [email protected].

Entity: LIVECONTEXT - Société par actions simplifiée (SAS)

Registered office: 173 rue de Courcelles, 75017 Paris, France

Share capital: 1 000 €

SIREN: 105 231 153

SIRET (head office): 105 231 153 00013

RCS: Paris 105 231 153

Legal representative: Lucky DONGANG

Email: [email protected]

Account information: Name, email address, username and password (stored only as a hash, never in readable form).

Profile data: Profile photo, preferences and usage history.

Technical data: IP address, browser type, operating system and general location.

Usage data: Conversations, workflows, agent interactions and platform activity.

Partner program data: If you apply to or take part in our partner program: the details of your application, the record of each acceptance of the Partner Program Terms (version, date and time, IP address and browser), and the identity, tax and bank details needed to pay your commissions.

Marketplace creator data: If you sell on the marketplace: the record of each acceptance of the Creator Terms (version, date and time, IP address and browser), your sales, royalties and payouts, and the identifier and status of the Stripe account through which we pay you. Stripe collects and verifies your identity and bank details itself, as an independent controller.

We use your data to:

• Provide and maintain our services
• Manage your account and subscriptions
• Improve platform performance and features
• Send important notifications about your account
• Ensure platform security and prevent fraud
• Comply with legal obligations

• Contract: To provide the services you signed up for
• Legitimate interest: To improve our services and ensure security
• Legitimate interest (evidence): To keep the proof of the contracts you accept, such as the Partner Program Terms and the Creator Terms, and to establish or defend legal claims
• Consent: For optional communications and newsletters
• Legal obligation: To comply with applicable regulations

We may share your data with:

• Infrastructure providers: For hosting and platform operation
• AI model providers: To process your requests with the model you select (see Sub-processors below for the list, and section 7 for the stricter rules that apply to Google Workspace data)
• Payment processors: To handle billing securely
• Authorities: When required by law

We never sell your personal data to third parties.

Limited Use disclosure. LiveContext's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. The use of raw or derived user data received from Google Workspace APIs adheres to the Google Workspace API User Data and Developer Policy, including the Limited Use requirements: it is never used, transferred or sold to develop, improve or train generalized AI or machine learning models.

LiveContext can connect to your Google account through these Google products: Gmail, Google Drive, Google Docs, Google Sheets, Google Slides, Google Forms, Google Calendar, Google Chat, Google Contacts, Google Tasks, Google Classroom, Google Workspace Admin (directory), YouTube, YouTube Analytics, Google Analytics, Google Ads, Search Console, Google Business Profile, Google Books, Perspective, and Google Cloud (BigQuery, Cloud Storage, Cloud SQL, Firestore, Pub/Sub, Cloud IAM, Cloud Run, Cloud Functions, Cloud Translation, Cloud Vision, Cloud Natural Language, Speech-to-Text, Vertex AI, Dialogflow). Some of them are available only through your own Google Cloud OAuth client (see Restricted scopes below).

If you connect a Google account to LiveContext, we access only the OAuth scopes you explicitly authorize during the Google consent flow. When you connect an integration, the consent screen asks for all the scopes that integration declares, not only those a particular workflow step will use, and we record the scopes Google actually granted.

Use. Google user data is used solely to execute the workflow operations the user has configured - for example, sending an email through your Gmail account, creating a Google Calendar event, appending a row to a Google Sheet, querying BigQuery, or generating a translation via the Cloud Translation API. Data accessed through Google APIs is never used for any purpose unrelated to executing the user-configured workflow.

Storage. OAuth refresh tokens are stored encrypted (AES-256) in our credential vault and are never displayed, not even to you. API responses are stored as the output of the workflow step or chat tool call that produced them, so you can inspect the run, and are tied to your account. Application logs record the names and sizes of the data a tool call or a model request carries, not its content.

Workspace sharing. A Google account you connect belongs to the workspace you connect it in. The other members of that workspace can use it in the workflows and AI agents they run there, although LiveContext never displays its tokens to them. Your personal workspace has no other members unless you invite them, so to keep a Google account private, connect it in your personal workspace.

Retention of Google Workspace data. Output derived from Google Workspace (Gmail, Google Drive, Google Docs, Google Sheets, Google Slides, Google Forms, Google Calendar, Google Chat, Google Contacts, Google Tasks, Google Classroom and Google Workspace Admin) is tagged as restricted when it is written and is excluded from message search: it is kept for at most 30 days. After 30 days it is deleted from storage: the workflow step outputs, the files that a workflow step or an AI agent produced from it, and the stored payloads that carried it. In chat conversations and agent execution logs it is redacted: the tool results that carry it, and the messages and tool-call arguments that quote it, are replaced by a short notice that the content was removed, and the conversation summaries built from it are cleared. Deleted content can remain in our encrypted backups for the periods stated in section 9.

What you build is not deleted automatically. The 30-day limit above does not apply to what you explicitly build in LiveContext with that data: tables (for example, a workflow that saves Gmail messages into a table you designed), agent tasks, and interfaces. They are storage you direct and control, the same way a spreadsheet you save data into is yours to keep, so they are not automatically deleted: they remain until you edit or delete them yourself, delete the table, task, interface or workflow that holds them, or until your personal workspace is deleted with your account (see section 10). Table rows and agent tasks written from Google Workspace content stay tagged as restricted and are never sent to a model provider outside the Anthropic/OpenAI allow-list described below (for the Workspace products other than Gmail and Google Drive, rows and tasks written from October 12, 2026).

Sharing. Google user data is never sold, never used for advertising, and never shared with third parties except as strictly required to execute the workflow the user configured (for example, calling the target Google API endpoint). We never transfer Google user data to other apps outside the scope of the user-configured workflow.

Human access. No one at LiveContext reads your Google user data, except: with your explicit permission for specific items (for example, the messages or files you point us to when you ask for support); when it is necessary for security purposes, such as investigating abuse or a security incident; to comply with applicable law; or for our internal operations, and then only after the data has been aggregated and anonymized. Our monitoring works on error types, counts and technical log records, never on the content of your Google user data.

AI processing of Google Workspace data. When you ask an AI agent to work on data from Google Workspace (any of the products listed in the retention paragraph above), that data is sent only to Anthropic and OpenAI models, through their direct APIs, whose API terms exclude API content from model training for every account. Requests to OpenAI carry the store: false flag. Google (Gemini API) and Mistral models are not used for this data: their training exclusion depends on the account tier or an account setting, which LiveContext cannot guarantee. If the selected model belongs to any other provider, including Google, Mistral, model aggregators and the command-line coding assistants, LiveContext refuses to send the data and asks you to switch model. This applies to the tool result itself, to the rest of that conversation (including its automatic summaries), to sub-agents working for that agent, and to later AI steps of the same workflow run (agent, classify, guardrail and browser agent steps, and child workflows it starts). A workflow run holding such data cannot create or update an agent task, because the model that will work on the task is not known in advance.

No AI/ML training on Google Workspace data. LiveContext does not use any data accessed via Google Workspace APIs (Gmail, Google Drive, Google Docs, Google Sheets, Google Slides, Google Forms, Google Calendar and the other Workspace products listed above) - including message content, headers, file content, file metadata, event data, sheet rows, or form responses - to develop, improve, or train generalized AI or machine learning models. Workspace data is processed only as needed to execute the workflow operations the user explicitly invokes, and is not retained for training purposes.

Revocation and deletion. You can revoke LiveContext's access at any time from your Google Account permissions page or by deleting the credential in LiveContext settings. When you delete the credential in LiveContext, we ask Google to revoke the access (unless another of your connections still relies on the same Google authorization, which revoking would cut off too) and delete the encrypted OAuth tokens immediately. When you revoke access from your Google Account instead, LiveContext detects it the next time it tries to refresh the access token: it then deletes the stored tokens and marks the connection as needing to be reconnected. When your account is purged (see section 10), we ask Google to revoke every access you granted and delete the tokens. Retained step outputs in your personal workspace are deleted with your account; those in a shared workspace stay with that workspace (see section 10). Any of them can be deleted on explicit request ([email protected]).

Restricted scopes. Google classifies the Gmail and Google Drive scopes that read your messages and files as restricted. LiveContext has asked Google to verify two of them for its own OAuth client: management of your Gmail messages without permanent deletion (gmail.modify), which includes reading them, and read-only Drive access (drive.readonly). Until Google approves them, LiveContext's OAuth client requests them only for the accounts used in Google's verification review, and for everyone else the features that need a restricted scope work only when you connect through your own Google Cloud OAuth client (BYOK) and grant the scope on your own consent screen; in that mode, the OAuth client is yours, not LiveContext's. LiveContext's OAuth client never requests full mailbox access (https://mail.google.com/) or full Drive access (drive): the features that need them, such as permanently deleting a message or editing a Drive file LiveContext did not create, work only through your own OAuth client, as do the other restricted scopes (read-only Gmail, drafts and mail settings).

Platform Terms. LiveContext's access to and use of data obtained through the Meta platform (Facebook Pages and Instagram professional accounts) adheres to the Meta Platform Terms and Developer Policies.

If you connect a Facebook Page or an Instagram professional account to LiveContext, we access only the permissions you explicitly authorize during the Facebook or Instagram login (OAuth) flow, and only for the accounts you connect. When you connect, the login asks for all the permissions the integration declares, not only those a particular workflow step will use.

Use. Meta platform data is used solely to execute the workflow operations you configure on your own connected accounts - for example, publishing a post to your Facebook Page or Instagram account, reading or moderating comments on your own posts, or reading engagement insights (reach, impressions) for your Page or posts. Data accessed through the Meta APIs is never used for any purpose unrelated to executing the workflow you configured.

Storage. OAuth access tokens are stored encrypted (AES-256) in our credential vault and are never displayed. As for Google accounts (see Workspace sharing in section 7), an account you connect belongs to the workspace you connect it in, and the other members of that workspace can use it in the workflows and AI agents they run there; connect it in your personal workspace to keep it private. API responses (account information, posts, comments, insights) are stored as the output of the workflow step that produced them, so you can inspect the run, and are tied to your account.

Sharing. Meta platform data is never sold, never used for advertising, and never used to develop, improve, or train AI or machine-learning models. It is never shared with third parties except as strictly required to execute the workflow you configured (for example, calling the target Meta API endpoint).

Revocation and deletion. You can revoke LiveContext's access at any time from your Facebook or Instagram settings (Settings & privacy → Business integrations / Apps and websites) or by deleting the credential in LiveContext settings. When you delete the credential in LiveContext, the encrypted OAuth tokens are deleted immediately. Access you remove in your Facebook or Instagram settings stops working at Meta at once; the token LiveContext still holds is deleted at the latest when you delete the credential or your account is purged. Retained step outputs in your personal workspace are deleted with your account; those in a shared workspace stay with that workspace (see section 10). Any of them can be deleted on explicit request ([email protected]).

• Account data: While your account is active + 3 years
• Usage data: 2 years
• Application logs: 7 days in our central log store and in the servers' system journals (they record field names and sizes, not the content of your data); container log files on the servers are rotated by size and overwritten
• Security audit logs: 90 days
• Google Workspace derived content (workflow step outputs, files and stored payloads; chat tool results and the messages quoting them; agent execution logs): 30 days, then deleted from storage, or redacted in chat conversations and agent execution logs (replaced by a notice that the content was removed)
• Google Workspace derived content saved into a table, an agent task or an interface you built: kept until you delete it or the workflow that holds it, or until your personal workspace is deleted with your account; it is storage you direct, so it is not automatically deleted. Table rows and agent tasks stay tagged as restricted and are never sent to a model provider outside the Anthropic/OpenAI allow-list (for the Workspace products other than Gmail and Google Drive, rows and tasks written from October 12, 2026)
• Billing records: 10 years (legal requirement)
• Partner program acceptance records: For the duration of your participation, then 5 years (the statutory limitation period), including after your account is deleted
• Partner commission, payment and tax records: 10 years (legal requirement)
• Creator Terms acceptance records: For as long as you sell on the marketplace, then 5 years (the statutory limitation period), including after your account is deleted
• Marketplace sales, royalty and payout records: 10 years (legal requirement)
• Deleted account: deactivated at once and restorable for 30 days, then purged (see section 10); the billing, partner program and marketplace records above are kept for their own periods
• Backups: deleting data does not remove it from our backups at once. Daily encrypted database backups are kept for up to 30 days, and a file you delete stays in the backup copy of stored files for up to 30 days. Weekly encrypted database copies kept with Backblaze are deleted automatically 90 days after they are made; they are used only to recover from a disaster.

Under the GDPR, you have the right to:

• Access your personal data
• Correct inaccurate data
• Delete your data
• Restrict processing
• Export your data (portability)
• Object to certain processing
• Withdraw consent at any time

To exercise any of these rights, email us at [email protected]. We will respond within 30 days.

Deleting your account. When you delete your account, it is deactivated immediately and you can restore it for 30 days; until then your data, including your connected credentials, is kept. After 30 days the account is purged: we ask each provider you connected (including Google) to revoke the access you granted, delete your stored credentials and tokens and your sign-in identity, and delete your personal workspace with what it holds (workflows and their runs, conversations, agents, tables, interfaces and files). A shared workspace you own is handed over to another member, and what you created in a shared workspace stays with that workspace, because its other members keep using it. The purge does not yet delete the direct messages you exchanged with other members, the logs of the calls received by your webhooks, forms and chat endpoints, your workflow variables, or the usage-analytics profile PostHog holds for your account; you can ask us to delete them ([email protected]). The records this policy says we must keep longer are kept for their own periods (see Data Retention).

We protect your data with:

• Encryption in transit: TLS 1.2 or higher between your browser and our edge, and from our edge to our servers. Traffic between our own services stays on a private network that is not reachable from the internet; it is not separately encrypted.
• Encryption at rest: stored credentials and OAuth tokens are encrypted with AES-256-GCM; stored files are kept in Cloudflare R2, which encrypts them at rest; database backups are encrypted before they leave our servers.
• Passwords stored only as hashes, and secure authentication mechanisms
• Continuous monitoring, continuous automated security testing (code analysis with CodeQL, dependency alerts with Dependabot, secret scanning with gitleaks) and internal security reviews. An external security assessment is planned.

We use cookies for:

• Essential: Required for the platform to work (authentication, preferences)
• Performance: To understand how the platform is used

You can manage cookies through your browser settings.

Our servers and databases are hosted within the EU (Germany). Stored files are kept in Cloudflare R2, which places them automatically and does not guarantee an EU location, and backup copies are kept with Backblaze (see Sub-processors). Some data may also be transferred outside the EU for AI model processing. These transfers are protected by standard contractual clauses approved by the European Commission.

Email: [email protected]

Postal address: LIVECONTEXT SAS, 173 rue de Courcelles, 75017 Paris, France

You also have the right to file a complaint with the CNIL (French data protection authority) if you believe your rights are not respected.

To deliver the platform, LiveContext relies on the following third-party processors. Each processor is bound by a data-processing agreement and either operates within the EU or provides Standard Contractual Clauses for any transfer outside the EU.

• Hetzner Online GmbH (Germany, EU) - Infrastructure hosting (compute, storage, databases, backups).
• Anthropic, OpenAI, Google, Mistral - LLM inference for AI-agent and assistant features, through their direct APIs. Customer prompts and outputs are sent only when the customer invokes an AI feature. Only Anthropic and OpenAI may receive Google Workspace data (see section 7).
• Stripe Payments Europe (Ireland, EU) - Subscription billing and payment processing, and payouts to marketplace creators through Stripe Connect (Stripe verifies the identity and bank details of creators as an independent controller).
• Resend - Transactional email delivery (account verification, contact form, operational notifications).
• Cloudflare - CDN, DDoS protection, and WAF for the public website and API edge.
• Cloudflare R2 (Cloudflare, Inc.) - Object storage for the files you upload or that your workflows produce, including Google Drive files and Gmail attachments a workflow step saves as a file, and our encrypted daily database backups. R2 chooses the storage location automatically, so these files are not guaranteed to stay within the EU.
• Backblaze B2 (Backblaze, Inc.) - Backup storage for disaster recovery: a weekly copy of our databases, encrypted before it leaves our servers, and a nightly copy of the stored files.
• PostHog (EU-hosted instance) - Product analytics: usage events such as which pages and features are used. Session recording is turned off, and PostHog receives no content of your chats, workflows or files and no Google user data.

Other model providers you can select. When you choose a model from DeepSeek, xAI, Perplexity, Cohere, Z.ai, Alibaba Cloud (Qwen), Moonshot AI, MiniMax or OpenRouter, the content of that request is sent to the provider you selected and is processed under that provider's own terms. OpenRouter is an aggregator that forwards the request to the upstream provider of the chosen model; requests to it ask for providers that do not store or train on prompts. The command-line coding assistants (Claude Code, Codex, Gemini CLI, Mistral Vibe, Cursor) available for some models run on the platform's own subscriptions. Google Workspace data is never sent to any of them.

Internal monitoring. Alerts and a twice-daily review of the production service are sent to our operators over Telegram, and the review is prepared with Claude Code (Anthropic) on our own subscription. These two services are used under their standard terms, without a data-processing agreement, so we limit what they receive: error types, counts and technical log records, which can include technical identifiers such as an account id, but never the content of your data and never your Google user data.

Self-hosted installs. Self-hosted installs using LiveContext cloud credits send their model requests through our relay to the provider chosen on that install. The relay forwards each request and its answer and keeps only the token usage needed for billing.

This list is updated when we onboard a new sub-processor. Material changes are announced by email at least 30 days before they take effect, giving customers an opportunity to object. To request a copy of the current list with contractual safeguards, email [email protected].