Privacy Policy

Who this covers

LEMAIT is operated as an early-stage, operator-run product (see About). A formal operating entity is being finalized; until it is, this policy describes how the LEMAIT platform itself actually handles data, and we'll update it the moment that changes — we won't invent entity details in the meantime.

What we collect

• Account data: your email and a bcrypt-hashed password — we never store or see your password in plain text.
• Task content: the descriptions you submit and any files/data you attach, so they can be estimated and, once you order and pay, worked on.
• Billing contact info: whatever's needed to send an invoice (currently handled manually by bank transfer — see Terms; we don't yet process card payments, so no card data touches our systems).
• Enquiries you send us: if you use a contact form on this site, we store the name, email, optional company and message you typed, so we can reply. That's the only thing we use it for — no newsletter, no profiling, no passing it on. Ask us and we'll delete it.
• Language preference: stored in your browser's local storage, not a tracking cookie — it just remembers which language you picked.
• A one-time IP lookup: on your first visit, your browser asks a third-party service (geojs.io) which country your IP belongs to, purely to guess your language automatically. We don't store this IP or the result — it's used in the moment and discarded. The notice shown on your first visit lets you switch to "English only," which stops this lookup from running on future visits.

How task content reaches AI models

Where processing physically happens, stated plainly: our current providers are Anthropic, Google, OpenAI and OpenRouter-hosted models — all US-based. None of them is an EU company. “EU AI Act mode” is therefore about control and traceability — which providers are eligible, and a per-task record of which model ran — not about processing inside the EU. If your use case requires EU-resident processing, tell us before you order and we will say plainly whether we can meet it.

Before a task description is sent to any AI provider, it passes through an automated redaction step (built on Microsoft Presidio). It detects and replaces with placeholders: email addresses, phone numbers (international and Russian formats), IBANs, EU VAT identifiers, Italian tax codes, French company identifiers, payment card numbers, and Russian tax/social identifiers. Bank, card and tax identifiers are additionally checksum-validated, so the detection is precise rather than approximate.

What it does not catch, stated plainly: personal names written in free text are not automatically redacted. Name detection requires a language model per language, and we do not currently run one we would trust across English, German, French and Italian. Rather than claim a protection we cannot demonstrate, we tell you this: if a name matters, write “the client” or “our finance lead” instead. Everything structured — how to reach someone, how to pay them, how to identify them officially — is removed automatically.

The task is then routed to whichever AI provider is capable of the work. If your workspace is in EU AI Act mode, providers we consider unsuitable for EU-regulated work are excluded from routing entirely — today that means models hosted by Chinese providers — and the specific model that handled each task is logged and visible to you. In the default (Free) mode, every provider is eligible; redaction applies either way.

Where data is stored

The application database runs on a server in Nuremberg, Germany (EU), reachable only over HTTPS and not exposed to the public internet directly. We don't use tracking cookies or third-party advertising pixels; the traffic-analytics tool on this site (Umami) is self-hosted on the same infrastructure and doesn't use cookies or collect personal identifiers.

Your rights

You can ask us, at any time, to see what we hold about you, correct it, or delete your account and associated data. Email ai@lemait.ai and we'll handle it personally — there's no ticket queue. We keep account and task data for as long as your account is active, and delete it on request except where we're required to keep billing records for a legal retention period.

What we don't do

• We don't sell your data, to anyone, ever.
• We don't run advertising or behavioral-tracking cookies on this site.
• We don't use your task content to train models — it's passed through to the AI provider handling that task and nowhere else.

Changes to this policy

If this policy changes in a way that affects how your data is handled, we'll update the date above and, for material changes, email active account holders directly.