Privacy Policy

1. Data Controller

The data controller responsible for your personal data is:

For privacy-related inquiries, please contact us at: legal@im-permanent.com. We have not appointed a data protection officer.

2. What This Notice Covers

This notice covers two things:

• This website – the public marketing site, including its contact form.
• The (IM)PERMANENT platform – the signed-in application, the admin portal and the pipeline portal, including the archive records, imagery, 3D capture processing and recommendations they provide.

Your use of the platform is also governed by our Terms of Service.

3. What Data We Collect

If you use the contact form on this website:

• Name
• Email address
• Company or role (optional)
• Your message content
• A record that you gave consent, which version of this notice you were shown, and when. That record stores a one-way hash of your email address rather than the address itself.

If you hold a platform account:

• Email address, password (stored only as a hash), display name, avatar URL, locale, last-seen time, session and refresh tokens, role grants, and your organisation memberships.
• The archive records, items, boards, notes, tags and shares you create, and the files you upload. Uploaded images keep their EXIF metadata, which can include GPS coordinates, capture times and camera serial numbers. We do not currently strip it.
• Full-resolution capture imagery submitted for 3D reconstruction, and the job records describing that processing.
• Behavioural events – which items you view and open, which you favourite, which boards you add them to, which recommendations you click, and the text of the searches you run. Each event is stored against your account with a timestamp.
• The recommendation rows we compute from those events, with the score and the reason for each.

Whenever anyone uses either surface:

• Operational logs. Our application log records a pseudonymous account identifier, the request path and the outcome – never your IP address, email, request body or headers. Our load balancer, CDN and network logs do record IP addresses, and content-security-policy reports record the referring page and browser user agent.

4. Why We Process It, and On What Legal Basis

• Running your account – registration, approval, authentication, organisation membership, password reset and email change. Basis: performance of a contract, Art. 6(1)(b).
• Providing the archive – creating, curating, illustrating and sharing records, items, boards and notes, and storing the files you upload. Basis: performance of a contract, Art. 6(1)(b).
• 3D capture processing – turning capture sets into Gaussian splat reconstructions and derived assets. Basis: Art. 6(1)(b).
• Automated classification of imagery and search queries – classifying garments, interpreting what you typed into search, and turning archive item text and your search text into search vectors so that search can match meaning. Basis: Art. 6(1)(b).
• Personalised recommendations – recording behavioural events and computing recommendations from them. Basis: our legitimate interests, Art. 6(1)(f) – specifically, making a large reference archive navigable so that the people who pay for it can find relevant pieces. We consider this a limited intrusion: the data is behaviour inside a professional research tool, it is never used to make any decision about you, it is never sold or shared for anyone else's purposes, and you can object at any time and we will stop – see section 10.
• Responding to your enquiry when you use the contact form on this website. Basis: your consent, Art. 6(1)(a), given by ticking the box on the form. Keeping the record that you consented is itself based on our legal obligation under Art. 7(1) to be able to demonstrate it.
• Operating, securing and administering the service – admin actions, fault diagnosis, abuse detection and the operational logs above. Basis: our legitimate interests, Art. 6(1)(f) in running a service that works and is not abused.

Where we rely on legitimate interests you may object under Art. 21 – see section 10. Where we rely on consent you may withdraw it at any time – see section 11.

5. Who Receives Your Data

We do not sell your personal data and we do not share it for anyone else's marketing. We do use service providers, and they receive personal data in order to provide those services. This is the complete list:

• Amazon Web Services – hosting, databases, file storage, search, logging and the email we send you about your account. Everything at rest sits in the Paris region (eu-west-3); the content-delivery network and the TLS certificate service run in the United States.
• Amazon Bedrock (an AWS service) – three models. Claude Haiku 4.5 interprets your search queries; Claude Sonnet 4.6 analyses garment imagery; Titan Text Embeddings V2 receives an archive item's title, description and confirmed labels (not the designer's name), and the text you type into archive search (up to 500 characters), to rank search results. A title or description is curator free text and may incidentally contain personal data. Your search text is not linked to your account when it is sent; for this step the text and the vector made from it are held in our server's memory for at most ten minutes and never written to storage. The searches you run are still recorded in your activity history, as section 3 describes. All of that text is sent in the Paris region (eu-west-3) only, and the model provider neither keeps it nor trains on it.
• Thunder Compute (United States) – GPU instances that receive the full-resolution capture images for a reconstruction job.
• Have I Been Pwned, via Cloudflare – when you set a password we send the first five characters of its SHA-1 hash to check it against known breaches. The password itself never leaves our systems.
• Google Workspace (United States) – hosts our mailboxes, so it receives any email you send us, including messages to legal@im-permanent.com and every contact-form submission.
• Hugging Face – listed for completeness. We download machine learning model weights from it. It receives no data about you.

We may also disclose personal data where we are legally required to do so.

6. Transfers Outside the EU

Our databases and file storage are in the European Union and are not replicated outside it. Several of the recipients above are not, and these transfers are real:

• Thunder Compute – the GPU provider does not tell us, and we cannot specify, which country a given instance runs in.
• Google Workspace – United States, for email you send us and for contact-form submissions.
• Content-delivery network access logs – US East. They record viewer IP address, URL and user agent.
• Have I Been Pwned – served from Cloudflare's global edge.

We do not currently have standard contractual clauses, an adequacy decision or a transfer impact assessment in place for these transfers. We are stating that plainly rather than implying a safeguard that does not exist. It is being worked on, and this section will be updated when it changes.

7. How Long We Keep It

We would rather tell you what actually happens than publish a retention schedule we do not enforce.

• Behavioural events – deleted after 90 days. This happens as part of an hourly maintenance job rather than as a separate scheduled task.
• Your account, content and uploaded files – kept until you delete them, or until you delete your account. No other time limit is currently defined.
• Recommendations – kept until they are recomputed, until you object, or until your account is deleted.
• Capture imagery and reconstruction job records – no time limit is currently defined.
• Contact-form messages – they live in our mailbox and with our email provider, neither of which is on a deletion schedule. The consent record we keep alongside them holds only a one-way hash of your email address, the date and the notice version, and is retained as evidence that consent was given.
• Operational logs – between 30 and 90 days depending on the log. Some database-level log groups currently have no expiry.

Establishing enforced retention periods for the entries above that have none is outstanding work, and we will update this section when they exist.

8. Automated Decision-Making and Profiling

We do carry out profiling, within the meaning of Art. 4(4), and the previous version of this notice was wrong to say otherwise.

For signed-in users we build a preference profile from your own behaviour – items you viewed, opened, searched for, favourited and added to boards – weight those signals, score the rest of the catalogue for similarity against them, and show you the highest-scoring items as recommendations. The score and the reason behind each recommendation are included if you ask us for a copy of your data.

We do not make any decision about you by automated means that produces legal effects or similarly significantly affects you, within the meaning of Art. 22. Accounts are approved by a person, not automatically. Nothing about your access, pricing or standing is decided by a model.

Automated classification is also applied to imagery – identifying garment attributes. That is applied to pictures of objects, not to people, and produces no decision about you.

9. Cookies and Similar Technologies

We set no cookies at all – not on this website, and not in the signed-in application. We run no analytics service, no advertising or tracking pixel, no session recording and no consent-management platform. We self-host our fonts so that loading a page does not call out to a font provider either.

The one thing we do store on your device is your session: when you are signed in, your session and refresh tokens are held in your browser's local storage, not in a cookie. That is what keeps you signed in from one page to the next.

The behavioural events described in this notice are not collected by cookies or by any tracking technology in your browser. They are records the application writes on our own servers while you use it, on the basis of our legitimate interests under Art. 6(1)(f), which is why the control over them is an objection (section 10) and not a cookie banner.

10. Your Rights

Under the GDPR you have the following rights over your personal data:

• Access (Art. 15) and portability (Art. 20) – you can request a copy of the data we hold about you in a machine-readable format. If you have an account you can do this yourself: open your profile and use Download my data.
• Rectification (Art. 16) – you can correct inaccurate data, from your profile or by writing to us.
• Erasure (Art. 17) – you can delete your account and the personal data attached to it. Open your profile and use Delete my account, or write to us. A small number of records survive deliberately and we will tell you which: a log recording that the deletion happened (holding a one-way hash of your email address and row counts, no contents), and any copyright or IP attestation you signed, which we retain under Art. 17(3).
• Restriction (Art. 18) – you can ask us to limit how we process your data while a dispute is resolved.
• Objection (Art. 21) – where we rely on legitimate interests you can object. For recommendations this is a switch, not a request: open your profile and turn off Personalised recommendations. From that moment we stop recording behavioural events for you, and we delete the events and the recommendations we already hold for you. The objection is enforced on our servers, so it applies even to activity your browser had not yet sent us. If you turn it back on later, we start recording again from that point; we do not reconstruct the gap.
• Withdrawal of consent (Art. 7(3)) – see section 11.
• Complaint (Art. 77) – see section 12.

To exercise any right that is not a control in the app, write to legal@im-permanent.com. We will respond within one month.

11. Withdrawing Consent

Where we process your data on the basis of consent – today, that is the contact form on this website – you may withdraw it at any time by writing to legal@im-permanent.com. Withdrawing consent does not affect the lawfulness of processing carried out before the withdrawal. We will keep the record that consent was given and withdrawn, because Art. 7(1) requires us to be able to demonstrate it.

12. Right to Lodge a Complaint

If you believe your data protection rights have been infringed, you may complain to a supervisory authority. In Belgium this is:

13. Changes to This Notice

We update this notice when what we do with personal data changes. Every version carries a version number and an effective date, and the consent we record when you use the contact form stores the version you were shown – so it stays possible to say what this page said at the moment you agreed to it.

Version 2026-10-22 – 22 October 2026

Amazon Bedrock gains a third model, Titan Text Embeddings V2, so that archive search can match what a query means as well as the words it uses. It receives an archive item's title, its description and its confirmed descriptive labels (not the designer's name). A title or description is free text written by our curators and may incidentally contain personal data, for example a person's name. It also receives the text you type into archive search (up to 500 characters), so that results can match what you meant. Your search text is not linked to your account when it is sent. The model turns it into a list of numbers; our server keeps the text and those numbers in memory for at most ten minutes, so that repeating a search does not send it again, and this step never writes either to storage. (The searches you run are still recorded in your activity history, as they already were: see the section on what data we collect.) All of this text is sent in the Paris region only, the model provider neither keeps it nor trains on it, and it is used only to rank search results. Nothing else about your data changes.

Version 2026-10-01 – 1 October 2026

Three services no longer receive anything, because the features that used them have been removed, and two statements about features that had already stopped are corrected. DuckDuckGo: our system used to send a garment's description to DuckDuckGo's image search in the United States to find reference pictures; that garment research feature is gone. Titan Multimodal Embeddings: garment imagery used to be sent to this Amazon Bedrock model to generate similarity embeddings; that has stopped. Titan Image Generator v2: imagery sent for object removal and in-painting used to be processed by this model in the US East region; object removal has been retired. All three are removed from the list of recipients, and Titan Image Generator v2 from the list of transfers outside the EU, and our servers no longer have permission to call either Titan model. Amazon Bedrock is still used, now with two models: Claude Haiku 4.5 for your search queries and Claude Sonnet 4.6 for garment imagery, both as before. Separately, the notice no longer says we remove backgrounds from imagery, because that feature was removed earlier, and no longer says a contact-form handler runs in the United States, because that handler was shut down on 14 September 2026 as the entry for that date records. Both statements described processing that had already stopped. Nothing else about your data changes.

Version 2026-09-14 – 14 September 2026

Contact-form submissions are no longer delivered through Resend. Until this date every submission was sent to Resend in the United States, and was processed and logged by a handler in the US East region before Resend received it. Both are gone: the message is now sent from our own servers in the European Union directly to our mailbox. Resend is removed from the list of recipients and from the list of transfers outside the EU, its credential has been deleted, and the US East handler has been shut down. The mailbox itself is still hosted by Google Workspace in the United States, which was already disclosed and is unchanged. Nothing else about your data changes.

Version 2026-09-12 – 12 September 2026

We no longer claim to set a cookie, because we do not set one. The previous notice said the signed-in application stored a small cookie remembering whether you had collapsed the navigation sidebar. It did not: the code that would have written it belonged to a version of the sidebar that was replaced long ago and was never running, which we confirmed by checking a real browser session and the published application code. Nothing about your data changes — this removes a description of something that was not happening. We now set no cookie anywhere, on this website or in the application. Your session is still kept in your browser's local storage, as before, and that is unchanged.

Version 2026-09-07 – 7 September 2026

Gravatar was switched off, so Automattic no longer receives anything. Until this date, displaying your profile picture sent a hash of your email address, your IP address and the page you were on to Automattic in the United States, every time an avatar appeared. That has stopped: the setting is disabled on the live service, and Automattic is removed from the list of recipients and from the list of transfers outside the EU. Nothing replaces it — where you have not set a picture, the app draws your initials instead. No other change.

Version 2026-08-31 – 31 August 2026

Complete rewrite. The notice now covers the signed-in platform as well as this website — accounts, uploaded imagery, 3D capture processing, behavioural events and recommendations — names every recipient and every transfer outside the EU, records a legal basis for each activity, and describes the objection control in the app. It corrects two statements in the previous notice that were untrue: that no data was transferred to third parties, and that no profiling took place.

Version 2025-01 – January 2025

The first notice. It covered the marketing contact form only, and stated that no data was shared with third parties and that no profiling took place. Both statements were incorrect: contact submissions were delivered through an email provider in the United States, and the signed-in platform computed recommendations from behavioural events. No copy of that notice was archived, so this entry describes it rather than reproducing it.