Hylaq Air
live · WebRTC peer-to-peer · zero carrier in path
Hylaq 6G · Live Demo

Two phones.
Zero carrier.

Open this on two phones. Phone A taps Start; gets a 4-letter mesh code. Phone B types the code (or scans the QR). Both phones connect peer-to-peer via WebRTC. Send messages, send files, watch live throughput. Once paired, every byte flows direct phone↔phone — Hylaq's server stops touching the data plane entirely.

Mesh

idle
Phone BJoin an existing mesh

Why this is 6G, not just “mesh wifi”

3GPP's 6G targets define the generation by four properties — none tied to a specific frequency band:

AI-native RAN
On-device ML decides which peer / radio / link layer for each packet. Apple Neural Engine routes; Android NNAPI routes.
Integrated sensing + comms
Radio doubles as sensor. Signal quality from your phone tells the mesh where coverage is thin; we route around weakness in real time.
Network-as-a-service
No SIM. No plan. No subscription. You pledge bytes; you consume bytes. Loadit settles per-byte. The network is a market.
Hyper-distributed core
No EPC, no AMF, no SMF, no UPF — the cellular-core boxes 5G inherited from 4G. Every Hylaq node IS the core. State lives in the routing graph.

Layers under the live demo above

L1
L0 · IDENTITY
@handle is the IMSI
Cellular identity today is a 15-digit IMSI burned into a SIM. Hylaq replaces it with your @handle — already cryptographically signed, already portable, already free.
L2
L1 · RADIO
Every radio your phone already has
iOS uses AWDL (Apple Wireless Direct Link, ~250 Mbps peer-to-peer). Android uses WiFi Aware + WiFi Direct. BLE is the always-on control plane. 60 GHz mmWave (Wi-Gig) for high-density urban. Sub-6 cellular is the bridge of last resort.
L3
L2 · MESH
The live demo above
The chat + file transfer above runs over WebRTC DataChannel — direct phone↔phone once handshake completes. Same primitive scales to a 50m soft-cell of phones gossipping routing tables every 10s.
L4
L3 · BRIDGE
Anyone with internet bridges the rest
Most phones in a soft-cell don't need to bridge the global internet — only one in the cell needs LTE/5G/wifi backhaul, and they relay packets for everyone else. Settlement Grid pays the bridge node per byte forwarded.
L5
L4 · INTELLIGENCE
On-device routing model
A 4MB Core ML model on iOS, NNAPI on Android. Inputs: peer signal strengths, battery levels, link latencies. Output: best next-hop. Trained on the global mesh's anonymised routing decisions — every phone gets smarter as the network grows.
L6
L5 · SETTLEMENT
Per-byte, per-millisecond settlement
Loadit Settlement Grid handles micropayments. Each forwarded packet attaches a signed receipt; receipts batch and settle every 60s.

Free for me forever

Year-3000 ask: build a global cellular network with no partnerships, no recurring infra cost, no FCC license fee. Math says yes:

$0/mo
spectrum
AWDL, WiFi Aware, BLE, 60 GHz Wi-Gig, CBRS GAA — all unlicensed.
$0/mo
basestations
Users supply the radios. Apple already amortised the cost when you bought the iPhone.
$0/mo
backhaul
Bridge nodes pay their own data plan. Hylaq pays them for forwarding others' bytes — net positive.
$~80/mo
cloud
One small Vercel + Postgres bill for /api/six-g/* and Settlement Grid. Entire opex for a global carrier.
6.8 B
potential nodes
Smartphones on Earth. Every one has the radios. Every one is a basestation in waiting.
50 m
cell radius
Reachable peer-to-peer with current radios in current iPhones. No new hardware, no tower.

Stack the rest of the carrier layer

Rename + Air install + 6G mesh = the full Hylaq carrier stack. Rename puts “Hylaq” in your status bar via open-source kfd; Air auto-joins nearby Hylaq hotspots; 6G runs the peer-to-peer mesh you just used above. Every layer bootstrapped, $0 to anyone.