DAY PASS
€3/day
The impulse buy. One box, one day, no subscription needed.
- 4 vCPU · 8 GB RAM · 80 GB NVMe
- 24 hours, billed once — no auto-renew
- Every agent CLI preinstalled
- rel4i connect (PWA) included
ALWAYS-ON · ONE PERSON, ONE BOX
A home for your coding agents.
An always-on cloud box for Claude Code, Codex, Gemini CLI or any agent — 4 vCPU / 8 GB / 80 GB NVMe, every CLI preinstalled, driven from your phone with rel4i connect. Bring your own subscription — we never see your tokens.
From €3/day. No subscription needed to start.
$ sign in → pay → running agent in ~90s
$ the phone here is real — three recorded sessions, tap around
▶ LIVE · REAL RECORDINGS
// SETUP
From payment to a running agent in under two minutes.
Sign in, pay, and your box is live. No wizard to fight, no infra to configure.
01
Sign in with GitHub
One click, no forms, no passwords to invent. Your GitHub identity is all we ask for.
$ sign in with github → no password to invent
02
Pay
Day Pass €3 or Month €19, straight to Stripe checkout. Payment happens before we ever provision anything.
$ pay → box provisioning starts
03
Your box is live in ~90s
Open the link, install rel4i connect to your home screen, and talk to your agent. That's the whole onboarding.
$ running agent in ~90s
// PRICING
One box. Two ways to pay.
Same specs either way — 4 vCPU / 8 GB RAM / 80 GB NVMe, every agent CLI preinstalled, never asleep.
MONTH
€19/month
For agents that live on your box full-time.
- 4 vCPU · 8 GB RAM · 80 GB NVMe
- Runs 24/7, renews monthly — cancel any time
- Every agent CLI preinstalled
- rel4i connect (PWA) included
ADD-ON · VOICE DICTATION
Stop typing on glass. Talk.
Live speech-to-text inside rel4i connect — your words stream onto the screen as you speak, first words in about a second, then a top-tier model tidies the ramble into a clean prompt before it reaches your agent.
- Your first hour (60 min) is on us — one time
- After that: 2.5¢/min, opt-in only — nothing bills without your consent
- Live streaming transcription + best-model cleanup pass
- Works in every tab of rel4i connect
Voice audio is transcribed under rel4i's speech provider key — the one feature that routes through us, which is exactly why it's opt-in and metered. Your agent tokens are untouched.
2.5¢/min
first hour free, one time · after that: 120 min ≈ €3
ROLLING OUT — LANDS IN YOUR DASHBOARD
GET A BOX, THEN ADD VOICE →
Your only other cost: your Claude/ChatGPT subscription (or start free with Gemini). We never bill you for tokens and never see one — you log into your own agent accounts inside your own box.
// PRIVATE BY DESIGN
We run the box. We never touch your keys.
Bring your own agent subscription and log into it inside your box — there's nothing on our side to leak. Every customer gets their own machine, not a shared tenant.
0
Tokens we ever see
You authenticate Claude, Codex or Gemini inside your own box. None of it ever routes through rel4i.
1
Box, only yours
One subscription, one human, one machine. No shared tenants, no team console.
∞
Agents welcome
Agent-agnostic by design — any official, unmodified CLI. Neutral is the whole point.
// WHAT'S TRUE TODAY
· your own VM — not a shared container. Root is yours.
· you log your Claude / Codex / Gemini account in on the box — we never proxy or store your tokens
· no inbound ports open, SSH off by default — reachable only through your private link
· roadmap: rel4i technically locks itself out of your box (encrypted volume, key only yours)
ALPHA HONESTY: full self-lockout is on the roadmap, not shipped yet — during the alpha rel4i keeps an access credential to operate/support your box. We never touch your agent tokens. Full write-up on the /privacy page.
// STATUS
Alpha today. Here's exactly what that means.
No spin — a plain list of what's real right now, and what's still on the roadmap for Beta.
What works today
- Real Hetzner cloud VPS per customer — 4 vCPU / 8 GB RAM / 80 GB NVMe, never a shared tenant
- GitHub sign-in + Stripe checkout — live payments, box provisioned only after payment settles
- Claude Code, Codex and Gemini CLI preinstalled, agent-agnostic by design
- rel4i connect (phone PWA) included — your own root/SSH access, inbound ports closed by default
- We never see your agent tokens — you log Claude, Codex or Gemini in yourself, inside your own box
- Voice dictation add-on rolling out into the dashboard
What's coming for Beta
- Full self-lockout — rel4i rotates away and deletes its own access credential at handover
- An encrypted workspace volume only you can unlock — even a provider-level snapshot sees ciphertext, not your data
- Safe park/resume — a box can pause instead of just being deleted when a plan ends
- Open, verifiable image builds, so trust in the base image is provable, not just promised
- A self-serve export/delete flow and a formal GDPR review of the privacy policy
- A proper legal review of the Terms as the Founding 50 launch scales into general availability
Until Beta ships: rel4i keeps an operating credential to your box, and boxes are ephemeral when a plan ends — no snapshot or restore yet. Treat every box as disposable and push/export what you want to keep. Full detail on the /privacy and /terms pages.
// FAQ
Questions, answered.
What exactly is "a box"?
A private cloud VPS — 4 vCPU, 8 GB RAM, 80 GB NVMe — provisioned just for you, with every supported agent CLI preinstalled and running 24/7.
Do I need an AI subscription?
Not to start. The box runs Gemini CLI free out of the box. For Claude Code or Codex, bring your own Claude/ChatGPT subscription — that's your only other cost.
What happens when my plan runs out?
During the alpha, boxes are ephemeral — when your plan ends, the box and its disk are deleted. Push your work to git or export anything you want to keep before then. Encrypted, resumable parking is coming; until then, treat every box as disposable.
Can I SSH into my box?
Yes — it's your box. SSH is off by default after the handover ceremony (see Private by Design), but you can re-enable it any time; there's no lock-in.
How does the voice dictation add-on work?
Your first hour (60 minutes) of live dictation in rel4i connect is on us — a one-time free allowance. After that it's 2.5¢/min, billed monthly — but nothing bills without your explicit consent; without it, dictation just pauses once the free hour is used up. Audio goes to our speech provider under rel4i's key (the only feature that routes through us), your agent tokens stay untouched. Rolling out now — it'll appear in your dashboard.
What should I expect from an alpha?
Rough edges. Things may break, change, or need a restart. We're building in public and fixing fast — tell us what hurts.
What does Alpha mean here — and what changes in Beta?
Alpha means real infrastructure and real payments, run by one person — with a few honest gaps we're not hiding: rel4i still holds an access credential to operate and support your box, and boxes are ephemeral with no snapshot or restore. Beta closes those gaps: full self-lockout (we delete our own credential at handover), an encrypted volume only you can unlock, safe park/resume instead of hard deletion, and a formal legal/GDPR review of the Terms and Privacy pages. See the status section above for the full list.
Will my box and data survive the Alpha → Beta transition?
The upgrade itself won't be a reason to wipe your box — it lives as long as your plan is active, same as always. What changes is the safety net for when a box does end: today that's kill-only (no snapshot), which is exactly what Beta's encrypted-volume park/resume is meant to fix. Until it ships, keep treating every box as disposable — push your work to git or export anything you want to keep.
Refunds?
Unused full months are refundable on request. Day Passes are a small, one-time charge and are non-refundable once the box has booted.
Which agent CLIs are supported?
Claude Code, Codex, and Gemini CLI ship preinstalled today. We're agent-agnostic — official, unmodified CLIs only.
Is my code or data private?
Your box is yours alone — no shared tenants. See the /privacy page for exactly what we store (and what we structurally can't see).