Security & Privacy Architecture

The short version

Two separate pieces make up Hire Alien. This website handles your account, your billing, and shows you how many messages your AI employee answered. A desktop app — which you download and run on your own computer — is where your documents live, where your AI provider key is entered, and where every actual conversation happens. Nothing in the second category ever reaches the first.

The exact data flow

What	Where it lives	Do we ever see it?
Your account (name, email, password)	Our servers	Yes — needed to run your account
Billing (Safepay subscription status)	Our servers + Safepay	Yes — needed to bill you correctly
Your AI employee's name, role, tone	Our servers	Yes — this is just an identity label, not private content
Message counts and response timing	Our servers	Yes — anonymous counts only, powers your live analytics
Your documents (knowledge folder)	Your own computer, in the app	No — never uploaded anywhere
Your OpenAI/Claude API key	Your own computer, in the app	No — entered directly in the app, never transmitted to us
Actual customer conversations	Your own computer, in the app	No — processed locally, sent only to your chosen AI provider

Why this is genuinely true, not just a promise

The desktop app is real, runnable software you (or your IT person) can inspect — it's not a black box. When it answers a customer, it reads your local documents directly off your computer's disk, builds the AI's instructions locally, and sends only that single question-and-answer exchange to whichever AI provider you personally connected (OpenAI, Anthropic, or a fully local model that never touches the internet at all). Our servers are never in that path — there's no code that routes conversation content through us, because the app was built without that connection existing in the first place.

What this means practically

If you use a fully local AI model (no internet connection at all), your customer conversations never leave your building — not to us, not to any third party.
If you connect a real OpenAI or Claude API key, that conversation goes only to that provider, under their own data-use terms — not through us as a middleman.
Deleting the app removes everything it processed. There's nothing sitting on our servers to also delete, because it was never sent here.

Other real protections in place

Passwords are one-way hashed (bcrypt) — not even we can see your actual password.
Every login is a signed session token, not a permanent key.
Phone and WhatsApp messages are verified as genuinely coming from Twilio using cryptographic signature checks, so no one can fake a call or message to your AI employee.
Rate limiting protects against abuse on every public endpoint, including the one your AI employee's chat runs on.

Full legal detail lives in our Privacy Policy. Questions about any of this — .