Privacy Policy

1. What we collect

Duchamp (the “Service”) is a library of content templates you hand to an AI. Browsing, copying and MCP use require no account. The Service processes the following.

What we collect now

• Studio Ad beta: we store your brand brief, scripts and hooks, voice and speed, generated audio and captions, request identifiers, status and monthly usage. Briefs are sent to OpenRouter and connected model providers for scripts; approved scripts, voice and speed are sent to ElevenLabs for speech. These HTTPS requests exclude account identifiers and emails. Avoid unnecessary personal information.
• Account (if you sign up): your email address (email and Google sign-in). Kakao sign-in provides only a nickname and profile image; a contact email and display name are collected separately inside the Service.
• Template copy records: when you copy a template or download it as a .md file, we record which template it was (template id, kind, channel, length and the copied step for reels), whether the record is a copy attempt, a browser-confirmed copy or a download initiation, whether it was counted toward the daily allowance, the time, and either your account id (if signed in) or the visitor cookie value described in section 6 (if not). The copied text is not sent. We do not store IP addresses in these records, and we delete them after 90 days.
• MCP call records: when an AI tool (Claude, ChatGPT, etc.) searches or retrieves templates through the public MCP endpoint (/api/mcp) without signing in, we record the tool name, the search query (or the retrieved template id), the task summary, where and with which tool the result will be used, the template ids returned, the processing mode and time, and the client name (User-Agent). We do not store IP addresses in these records, and we delete them after 90 days.
• Usage events: when certain actions happen on screen (opening a template, pressing a copy button, etc.), we store in our database the event name, the page address, values attached to the action (such as a template id) and the time, together with your account id (if signed in) or the visitor identifier created by PostHog (if not).
• Usage analytics: through Google Analytics and PostHog we collect the pages you visit, the referring page, browser and device information, and the identifiers those tools create. We do not send your email or account details to them, and session recording and automatic click capture are turned off.
• Technical data: access logs, browser type and error records.

Kept only in your browser, never sent to our servers

• Recently viewed templates: for “continue browsing”, the ids and times of templates you recently opened are kept only in this browser's storage (up to 12).
• List you were browsing: when you open a template from a list, the list address to return to (including the chosen channel, type, tags, search words and sort), the detail address you opened, the scroll position and the time are kept in this tab only. The “← list” link and the sidebar use it to take you back to that list with the same filters. Records older than 30 minutes are not used, and they are removed when you close the tab.

Data retained from discontinued features

As of 26 September 2026 we no longer provide brand profiles, content generation, scheduling and publishing, channel connections, performance tracking, reference browsing and saving, the pre-sign-up preview, the free generator tools or pre-sales diagnoses, and we no longer collect the data below. Data you entered or connected before then remains in your account, so we still hold it; you may ask us to delete it at any time (section 5).

• Brand information: the brand site and channel URLs you submitted, public text read from those pages, and the brand profile and tone settings you confirmed or edited.
• Reference material: for public posts you added or saved — the URL, author, publication time, text read from captions and images, media URLs and public engagement counts — plus per-brand save lists, board names, your notes and the time they were saved.
• Output and records: generated concepts and drafts, your choices and edit reasons, scheduling details, and post URLs you registered with their public metrics.
• Connected channels (Threads, Instagram): only if you connected an account yourself through Meta sign-in — its username and profile picture, your own posts and their text, per-post insights (views, reach, likes, replies, reposts, quotes, saves, shares) and follower count, the addresses of posts published through the Service, and the access token. Tokens are stored in a server-only area and are never shown on screen or to other users. We no longer sync or publish with these tokens.
• Replies and mentions on a connected Threads account: the former “audience response” screen only displayed replies and mentions without storing them, so we hold none.
• Free tool and preview usage records: kept to rate-limit the former free generator tools and pre-sign-up preview — a hash of the IP address, a hash of the submitted URL, the tool name and outcome, the time, and your account id if you were signed in.

2. Why we use it

• Sign-in and account maintenance
• Providing templates, returning MCP search results and improving search quality
• Template usage statistics and counting free daily copies (when a daily limit applies)
• Providing paid subscriptions and confirming payment (once payments open)
• Holding data retained from discontinued features and handling export and deletion requests
• Operating the Service reliably, handling errors, preventing abuse
• Sending service messages (sign-in links, notices of important changes such as pricing or terms)

When an example image or video on a template is displayed, your browser may connect to the external platform's media servers, and your IP address and browser information are transmitted to that operator.

We do not sell collected information or transfer it to third parties for advertising.

3. Processors and third parties

The following providers process data to operate the Service. Each follows its own privacy policy.

• Supabase — database, authentication, file storage (Seoul region)
• Vercel — web hosting and server execution (Seoul region)
• Amazon Web Services (SES) — sign-in link email delivery (Seoul region)
• OpenRouter and connected model providers — when MCP search ranks results more precisely (deep search), we send the search query and task summary (300 characters combined at most) with candidate template details to score relevance. While the discontinued features were offered, they were also used for brand profiles, concept generation and reply summaries. We do not send your email address or account identifiers.
• TikHub — used by the operator to research public post data that templates are based on. No data about users of the Service is sent.
• Google (Google Analytics), PostHog — visit and usage analytics
• Google, Kakao — social sign-in
• Meta (Threads, Instagram) — if you connected a channel earlier, we hold the access token Meta issued for that connection. We no longer make requests to Meta with it.
• Paddle — payment processing for paid subscriptions as Merchant of Record, once payments open. Card details are entered on Paddle's systems and are never received or stored by us. Payments are not open yet, so no data is sent to Paddle today.

International transfers

Data is transferred outside Korea as follows to provide the Service. If you do not want a transfer, you can avoid the feature concerned (MCP deep search, paid subscription) or ask us to delete your account, in which case that feature is unavailable. Analytics data is not sent if you block the scripts in your browser.

• OpenRouter, Inc. (United States, openrouter.ai) and the model providers handling each request (United States and elsewhere) — data: the MCP deep-search query and task summary (300 characters combined at most) and candidate template details; no email or account identifiers. When and how: each time an AI tool requests a deep search, over an encrypted connection (HTTPS). Purpose: scoring search relevance. Retention: per each provider's policy after processing. While the discontinued features were offered, brand information, post text, images and, when a reply summary was requested, reply text (without usernames) were transferred the same way.
• Eleven Labs Inc. (United States and elsewhere, elevenlabs.io) — approved script, voice and speed are sent over HTTPS when you request speech to create narration and caption timing. Retention follows its policies and contract terms: privacy policy. Studio Ad briefs are also sent over HTTPS to OpenRouter and connected model providers when you request scripts.
• Google LLC (United States, Google Analytics) and PostHog, Inc. (United States) — data: pages visited, referring page, browser and device information, each tool's cookie identifiers and your IP address (processed by each provider). When and how: sent directly by your browser to each provider when a page opens (HTTPS). Purpose: visit and usage analytics. Retention: per each provider's policy and settings. Blocking these cookies or scripts stops the transfer and does not affect use of the Service.
• Paddle.com Market Limited (United Kingdom, Merchant of Record) — only once payments open. Data: email and payment and billing details. When and how: entered directly in Paddle's checkout when you pay (HTTPS). Purpose: subscription payment, refunds and tax. Retention: per applicable law and Paddle's policy.

4. Retention

• Studio Ad: kept while your account exists. Deleting a production record removes its input, scripts, stored audio and captions. Monthly usage and content-free request identifiers, hashes and status remain for quota accounting and duplicate prevention. Provider records follow their policies and contract terms.
• Account and data retained from discontinued features: kept while your account exists. If you delete your account, brand information, reference save lists, output and records are deleted within 30 days. If you ask us to disconnect a channel (contact below), we delete that channel's token and synced posts and insights; deleting the brand also deletes them. You can also remove the app connection in your Meta account settings.
• Template copy records and MCP call records: deleted automatically after 90 days. Deleting your account also deletes that account's copy records.
• Usage events and free tool usage records: kept for statistics while the Service operates. Deleting your account removes the link between those events and your account id; on request, we delete records linked to your account.
• Values stored in your browser: not held on our servers. Recently viewed templates remain in that browser until you clear them; the list you were browsing is removed when you close the tab.
• Access logs that the law requires us to keep are retained separately for that period only (up to one year). Once payments open, transaction records required by the Korean E-Commerce Act and other laws may be kept for the statutory period.

5. Your rights

• You may request access to, correction of, deletion of, or suspension of processing of your data.
• Requests to delete your account, or to export or delete data retained from discontinued features (brand information, output, connected-channel data and tokens, etc.), sent to sean@notique.co are actioned within 7 days.
• Users under 14 may not register.

6. Cookies and browser storage

The Service uses the following cookies and browser storage.

• Sign-in cookies: keep you signed in. Sign-in does not work if they are blocked.
• Visitor cookie (dc_vid): a random value our server issues the first time you copy or download a template without signing in. It is used to count free daily copies and for template usage statistics. It cannot be read by page scripts (HttpOnly) and lasts one year so the same browser is not issued a new value; copy records made with it are deleted after 90 days. Records from before and after you sign in are not merged. If you block or delete it, copying still works; while it is blocked, your copies are not counted.
• Analytics cookies: Google Analytics (cookies starting with _ga) and PostHog (cookies starting with ph_ and browser storage) use these for visit statistics. Blocking them does not affect use of the Service.
• Browser storage (localStorage): keeps recently viewed templates (duchamp:recent-views:v1) in this browser only (section 1). Not sent to our servers; kept until you clear it.
• Tab storage (sessionStorage): keeps the list you were browsing (duchamp.discovery-origin.v1, plus a temporary copy duchamp.discovery-restore.v1 used to restore the scroll position when you go back) in this tab only (section 1). Not sent to our servers; removed when you close the tab.

The Service itself does not set cookies for showing ads. You can block or delete cookies and storage in your browser settings.

7. Security

We apply transport encryption (HTTPS), row-level access separation in the database, and separate storage of secrets. We do not store API keys or passwords (sign-in is passwordless).

8. Changes

Changes are posted on this page with an effective date; significant changes are emailed to signed-in users.

9. Data protection officer

Sihyun Kim (Notique Inc.) · sean@notique.co