How Drop's AI works
Draft pending legal review
Entity details pending
Version 0.1 · updated on October 22, 2026
On an artist's page, whoever answers you in the chat is an artificial intelligence agent working on their behalf. This is what it does, what it does not do and what happens to what you tell it.
Who answers you
The agent talks to you on the artist's behalf: it gathers your idea, proposes slots from their schedule and prepares the project. The agent only holds slots: each appointment is confirmed or declined by the artist. The artist sees the whole conversation. When the agent cannot solve something, or you ask, the conversation goes to the artist, we tell you, and the agent stops answering.
What it does not do
It gives no medical or health advice: if you have a health question it recommends asking a professional and, in an emergency, calling 112, and it tells the artist. It charges nothing, confirms no appointments and does not talk about topics unrelated to your tattoo. If the artist asks for a deposit, the agent explains the amount, the deadline and the cancellation policy and sends you the payment link again, but it never treats it as paid and does not lower or waive it: the artist decides that. Nor does it read your consent or what you write in it. It can make mistakes: what counts is what the artist confirms.
Which models it uses
The agent uses models from OpenAI and Anthropic, reached through Vercel AI Gateway. GPT-6 Luna, from OpenAI, answers you, writes the summaries the artist asks for of a conversation or a project and the ones the agent uses as memory of long conversations, and reviews new messages to detect spam, harassment or threats. If it fails, a Claude model from Anthropic answers, and another Claude model reviews the messages. When moderation is on, an OpenAI model reviews the images sent to the chat. The full list of providers is in sub-processors.
What happens to what you write
The messages and images you send make up the conversation with that artist, and only they and you see it. They are used to serve you and prepare your project. The conversation reaches the AI providers, which process it in the United States with the safeguards explained in the privacy policy, for four purposes: answering you on the artist's behalf, writing the summaries the artist asks for, detecting abuse and, when it is on, moderating chat images. Under their business terms, they do not use it to train their models.
The first three only run while the artist's AI permission is on. Image moderation is a Drop safety measure and does not depend on the AI permission: it reviews the photos you send even when the agent is off.
Do not share health data in the chat: if you write it, it is removed from what the agent reads before it reaches the model, and what the artist needs to know is collected in the consent, which the agent does not read.
You can ask the artist for a copy or the deletion of your conversation, or write to legal@drop.ink from the email you use for the chat.
If something goes wrong
If a message looks like spam, harassment or a threat, the conversation may be paused and handed over to the artist, who reviews it. When image moderation is on, a sensitive image is blurred and an illegal one removed. If you believe it got it wrong, tell the artist or write to contact@drop.ink.
Entity details
- Trade name
- Drop
- Owner
- Pol Pages Luque
- Legal form
- Sole trader (autónomo)
- Tax ID
- 54165473J
- Registered address
- Avinguda de Catalunya 101, 08184 Palau-solità i Plegamans (Barcelona)
- Contact email
- contact@drop.ink
- Privacy email
- legal@drop.ink
- In force since
- [To be confirmed: In force since]