What the sandbox does with your data
Applies to the try-it sandbox only · Last updated 24 July 2026
Your sign-in email
We ask for your email to send you a one-time sign-in link — no password. We use that address only to:
- send you the sign-in link;
- keep the sandbox fair — a handful of runs and links per person per day, so one visitor can't exhaust it for everyone.
To do that we store your email address alongside each sign-in link we issue, together with the time and the requesting IP address (used only to enforce those fair-use limits). The link itself is single-use and short-lived; we store only a one-way hash of it, never the link, so a copy of our database could not be used to sign in as you. Your sandbox workspace is named by a one-way hash of your email, so your address does not appear in it. Signing in sets a signed cookie in your browser — there is no server-side session store.
The email you paste
When you paste a sample email, we run it through the same production pipeline a real mailbox would: the agent classifies it, grounds a reply in whatever knowledge is available, drafts a response, self-checks the figures, and routes it — then stops. Nothing is sent anywhere, because the sandbox has no mailbox and no send capability.
The pasted text, the generated draft, and a step-by-step audit trail of the run are stored in your sandbox workspace so we can show you the result and operate the demo. Paste sample or fictional content rather than anything sensitive — it's a demo. You can ask us to delete your sandbox data at any time at info@bydaniel.dev.
Where the AI runs — stated plainly
We do not claim that all your data stays in the EU. The sandbox itself is hosted and operated in the EU (Amsterdam). But the AI model that reads your email and writes the draft runs on Anthropic's infrastructure in the United States — there is no EU inference region on their first-party API. That transfer takes place under a data-processing agreement and EU Standard Contractual Clauses. Only the content needed to handle the request is sent, Anthropic does not train on it, and it is deleted on a 30-day standard.
AI transparency
Every draft the sandbox writes carries an AI-authorship line, in line with the EU AI Act's Art 50 transparency duty (applicable from 2 August 2026). The agent is built to escalate rather than invent facts: if it can't ground an answer, it says so instead of making one up.
Your rights & contact
Under the GDPR you can ask us to access, correct, or delete the personal data we hold about your use of the sandbox. Email info@bydaniel.dev and we'll take care of it. You also have the right to lodge a complaint with the Dutch data protection authority, the Autoriteit Persoonsgegevens (autoriteitpersoonsgegevens.nl), or the supervisory authority where you live.