Universal Reachability Layer

One address.

Your home server, NAS, smart devices, and AI agents — all reachable at one address you actually own. No technical setup. No subscriptions to chain together. Just text your server like a contact.

home-agent
my-home.airdr.es
Good morning! Ready to help. 08:59
What's my NAS storage at? 09:00
Used 1.4 TB of 4 TB — 35% full. 09:00
Type a message
Helena
helena · just now
skate park saturday? 🛹 21:13
the new one downtown 21:13
Riverside Skate Park
42 Harbor St · 0.4 mi away
21:13
project-agent
build.airdr.es
Phase 1 complete. 142 tests passing. Ready when you are. 14:02
Type a message
Personal Software

Software that belongs to you.

Run the tools and AI that matter to you on hardware you own — with the convenience you expect from the cloud.

01

Yours by default

Your data and your tools live on your hardware.

02

Tuned to your context

Personal AI works best with the things only you know — files, habits, projects.

03

Predictable cost

A box on your shelf, a few watts of power. No per-seat or per-token meters.

04

Yours to extend

Mix in any model, service, or cloud — connected the way you want.

The Problem

Hundreds of thousands of developers own compute
they cannot reliably reach.

The workaround stack is fragile, fragmented, and requires expertise that even experienced developers find tedious. Every solution solves one part — none solve all of it.

01 / Cost

Price floor

Hosted solutions start at €5/month minimum — before you add a database, storage, or a second service.

02 / Reachability

NAT & CGNAT dead ends

Self-hosted and P2P approaches silently fail behind carrier NAT, corporate firewalls, and CGNAT — with no fallback.

03 / Fragmentation

Five tools, one job

Nginx + Let's Encrypt + Docker + DDNS + Tailscale. Every piece breaks differently. None of them talk to each other.

04 / Sovereignty

Paths you can't audit

Many tunnels and storage layers are closed by design — you can't audit the path your traffic takes or verify the data at rest.

05 / Compute

Stranded hardware

Powerful machines — including GPUs — sit idle because exposing them remotely is more friction than it's worth.

06 / Cost creep

Compounding complexity

Stitching services together compounds — each addition adds a bill, a config surface, and a failure mode.

The Solution

One Airdress.
A permanent, routable identity for your infrastructure.

Airdress is an address layer. It sits in front of your devices and intelligently routes traffic to wherever your compute is actually running.

  • 01

    Multi-device routing with availability awareness

    Traffic automatically routes to your primary device. If it goes offline, failover to a standby happens within 5 seconds — no DNS TTL wait, no manual intervention.

  • 02

    Unified address across local and cloud compute

    One hostname. Behind it: your laptop, your home server, a VPS, or all three. Airdress resolves the right one at request time.

  • 03

    Sub-€1 always-on relay hosting

    A lightweight relay written in Rust. No container orchestration, no microservice mesh. Boring infrastructure that stays on without burning a budget.

A
my-agent
my-agent.airdr.es
Good morning! Ready to help.
What's my NAS storage at?
Used 1.4 TB of 4 TB — 35% full.
Type a message
Use Cases

Starting narrow. Expanding with every user.

The first wedge is webhook relays and personal AI agents. The same address layer scales to anything you want to reach over the internet.

Webhook relay

Stable URL for Stripe, GitHub, Slack — even when your laptop sleeps.

Personal AI agents

Reach your home model from anywhere. One address, end-to-end encrypted.

Homelab & self-host

NAS, media server, smart home — all reachable without exposing your IP.

IoT & edge devices

Sensors, gateways, ARM boards — addressable through CGNAT and dynamic IPs.

Preview environments

Per-branch URLs that route to your dev box. No tunnel restarts on every deploy.

Small-business infra

Run on-prem services with the reachability of a public service. Stable address, data on hardware you control.

Open Core

Open where it matters. Sustainable everywhere else.

The protocol, the relay, and the reference implementations are open source. The hosted control plane and the managed relay network fund the work.

Open source
  • · Protocol specifications
  • · Relay node (Rust, MIT)
  • · Client SDKs (Go, TypeScript, Rust)
  • · Reference WireGuard configurations
Hosted
  • · Managed control plane
  • · Global relay network
  • · Custom domain support
  • · Team accounts & audit logs
Design Principles

Built to ship, designed to last.

01 · Boring infra

Predictable over clever

Every component is replaceable, debuggable, and well-understood. No magic, no novelty for novelty's sake.

02 · Edge-first

Your hardware, your data

Compute happens where you want it — your laptop, your home server, your phone. Relays never see plaintext.

03 · Open by default

Verifiable, not trust-me

Source-available where it makes sense. Specs are public. Anyone can run their own relays.