How we use AI
What the AI writes
The personal text of each Page (the small label above the name, the headline, one to three sentences and the closing line) is written by a large language model provided by Anthropic (Claude). At the time of writing we use the model claude-sonnet-5. The animation, layout and design are hand-made templates, not AI output.
What the model receives
The occasion, the Recipient's first name, your name, the relationship and the details you type into the order form. It does not receive your e-mail address or payment details. We never send photos, because the Service does not accept any.
You decide what goes out
The generated text is shown to you as a watermarked preview before you pay. AI text can be inaccurate or awkward, so please read it. If a line is off, go back, adjust the details and preview again. If the model cannot produce usable text, a plain, pre-written greeting for the occasion is used instead, and the preview shows exactly what the Recipient will see.
Training and retention
Under Anthropic's commercial terms the model is not trained on your inputs or the generated text, and Anthropic deletes inputs and outputs within 30 days unless they are flagged for abuse. We do not create AI images, audio or video, and we do not generate content that imitates a real person's voice or likeness.
How the AI text is marked
Every Page carries a machine-readable generator tag in its HTML head stating that its text is AI-assisted, and every public page of this site links to this disclosure. This is provided in line with the transparency principles of Regulation (EU) 2024/1689 (the EU AI Act) and consumer transparency rules; we will update the marking when the Commission's code of practice on marking AI-generated content is finalised.
Questions: hello@sendawick.com.