Privacy, editorial, legal
What this site does with your data, how these posts are made and who answers for them. Written to be read, not to be survived.
Last reviewed: 27 August 2026
Privacy
This is a one-person studio site. There is no account, no newsletter and no advertising on it. One page has a form — the project enquiry on /ai — and everything it collects is described below. What follows is the whole of what happens to your information, not a summary of it.
Who is responsible
Ilona Golman, in Barcelona. For anything on this page, including any request about your data, write to ilonagolman@gmail.com.
What the site counts
Every page view sends three things to a visit counter: the hostname you are on, the path you opened, and the hostname of the site you arrived from — only on the first page of a visit. No name, no cookie, no identifier, nothing that follows you between visits.
Pressing certain things adds a second, separate record: the same hostname and path, a short fixed code for what was pressed, and whether your browser announces itself as automated. Never the text you typed, never anything you put in a form. The codes are fixed and look like this: share_x, out_linkedin, res_page-skeleton, mod_media, cta_whatsapp, wa_instead, wa_no_contact, bld_community, pick_school, more_building, inquiry_send. What carries one: a link that takes you off this site — a project, a resource, one of my profiles in the footer; the share row under an article; a Thalamus module on its page; the buttons at the top of a project page, including the one that opens WhatsApp; the send button on the enquiry form and the two ways out of it into WhatsApp instead; opening a card under Now building and following it through to the project; and expanding the hidden cards on a phone. A contact link that carries no code of its own — WhatsApp, email or phone — is recorded as the channel alone, never as the address it opens. Moving between pages here is not counted twice; the visit counter above already covers that.
Both counters are hosted on Supabase, which acts as the processor. Your IP address is not written into the record, but Supabase necessarily sees it while receiving the request, the same way any server sees the address it answers. Their servers may sit outside the EU.
The lawful basis is legitimate interest — knowing whether anyone reads this, and whether anything on the page is worth pressing. The interest is narrow and the data is minimal, which is why the counters were built without identifiers in the first place. Records are deleted after twelve months. What outlives them is a daily total per site with no paths in it, and that goes after twenty-four.
What is stored on your device
No cookies. Two preferences live in your browser's local storage and never leave it: the light or dark theme you picked, and the language and reading theme you last used on an article. Clearing your browser data removes them.
Video
Articles embed the video they discuss. The player does not load until you click it, and when it does it loads through youtube-nocookie.com, which does not set tracking cookies for a session you never play. The still image you see before clicking is served by Google, so opening an article with a video tells Google your IP address even if you never press play.
Fonts and other outside requests
Fonts are served from this site, not from Google. Apart from the video thumbnail above and the visit counter, the pages make no outside requests.
The project enquiry form
The form on /ai asks six questions. Five are optional; the last one — how to reach you — is required, because without it there is no way to reply. It collects only what you type: what you would hand to an agent, the stage, team size, timing, budget, and whichever contact detail you choose to leave. No cookie is set, nothing is stored in your browser, and the page does not know who you are before or after you send it.
When you press send, the answers travel to me three ways at once: as a message through Telegram, as an email through Brevo, which delivers it to my Gmail inbox, and as a card in the CRM I keep for my own work. Telegram, Brevo and Google each act as processors and each sees the contents in passing; the CRM is hosted by Supabase, which stores that card until I delete it. The card is the copy I actually work from — the site itself keeps nothing.
The lawful basis is your request: you are asking me to get in touch, which is a step taken at your initiative before any contract. I keep the enquiry as long as the conversation is alive, and delete it when it plainly is not.
The form also has a rate cap: no more than five enquiries an hour from one address. To count them the server takes your IP, mixes in a secret string and hashes it; only that fingerprint reaches the database — the address itself is never stored, and it cannot be recovered from the fingerprint. The counter row lives for an hour and is then erased. The lawful basis here is legitimate interest: without a cap a bot floods the form and a real enquiry is lost in the noise.
If you write to me
Email is email: whatever you send arrives with your address attached and stays in my mailbox for as long as the conversation is useful. Ask me to delete it and I will.
Your rights
You can ask to see, correct, delete or export what I hold, object to the processing, or withdraw a consent you gave. Write to the address above. If you think I have handled it badly, you can complain to the Spanish data protection authority, the AEPD.
Editorial responsibility
This section exists because Article 50(4) of the EU AI Act requires it of anyone publishing to inform the public on matters of public interest. It is a statement of how these pieces are made and who answers for them.
Who holds it
Ilona Golman holds editorial responsibility for everything published on this site. Nothing appears here that I have not read, checked and decided to publish.
How a post is made
- I watch or read the source in full — usually a long interview, a film or a policy document.
- I write an original summary and my own commentary. The piece is my argument about the source, not a restatement of it.
- I pull the claims that carry the argument and check each one against primary sources — the report, the filing, the transcript, the register.
- The check is published beside the piece, claim by claim, with a verdict and the sources behind it. You can disagree with me using my own evidence.
- Quotations are short, attributed to the speaker and marked with the timestamp they came from. Full transcripts are not published.
Where AI sits in that
I use AI tools throughout: transcription, research, drafting and translation into the four languages this site publishes in. They are instruments, and they are not the author. Every published piece has been reviewed by me on its substance — the argument, the facts, the wording — before it goes up, and I carry the legal responsibility for the result.
Corrections
If something here is wrong, write to ilonagolman@gmail.com. I would rather correct a piece than defend it.
Legal notice
Identifying information required of information society services under article 10 of Spanish Law 34/2002 (LSSI-CE).
- Holder
- Ilona Golman
- Tax number (NIF/NIE)
- —
- Address
- —
- ilonagolman@gmail.com
- Place of business
- Barcelona, Spain
- Activity
- Design, editorial and AI consulting; one-to-one AI sessions
- Website
- ilonagolman.com
Services sold from this site
One-to-one AI sessions: €80 in person in Barcelona or Sitges, €50 online. Booking runs over WhatsApp; payment for in-person sessions is in cash, after the session. There is no checkout on this site and no card details pass through it.
Content and reuse
The texts here are mine. Quote them with attribution and a link. Material quoted from other works — video excerpts, reports, documents — belongs to its authors and is used for review and commentary, attributed and sourced.
Applicable law
Spanish law applies. Any dispute goes to the courts of Barcelona, without prejudice to the rights a consumer has to bring a claim where they live.