01
Connect your computer
Contributors opt in, choose resource limits, and keep control. CPU-only computers are welcome.
THE NEXT CHAPTER OF COMPUTE LOCAL MVP
AI Compute.
Powered by People.
Turn the world's unused computing power into accessible AI infrastructure.
Explicit consent. Verified work. One unified API.
Independent devices. Shared possibility.Architecture illustration · not live network data
CPU-friendly by design One developer API Approved workloads only Rewards for verified work
A DIFFERENT KIND OF CLOUDSmall contributions.
Small contributions.
Collective intelligence.
Independent tasks, intelligently coordinated.
Built for useful AI work across everyday computers.
02
Put idle power to work
Our scheduler matches independent embedding tasks to compatible devices with time-bound leases.
03
Create shared value
Businesses receive verified results. Contributors receive simulated rewards for accepted work.
BUILT FOR DEVELOPERSYour AI workflow.
Your AI workflow.
A new compute layer.
Start with real text embeddings, powered by an approved MiniLM model. Submit a batch, track its progress, and retrieve normalized vectors.
Explore the APIyour-first-job.pyREST API v1
import httpx
# Public or synthetic data only
job = httpx.post(
f"{API}/api/v1/jobs",
headers=auth_headers,
json={
"workload": "text-embeddings",
"texts": ["Hello, connected world."],
"data_classification": "synthetic"
}
).json()TRUST STARTS WITH CLEAR BOUNDARIES
Read the trust model Your computer. Your choice.
No arbitrary code. No inbound ports. No contribution without permission. Results are checked before rewards are credited.
Consumer PCs are not confidential computing environments. This MVP accepts only public and synthetic data. All billing and rewards are simulated.
TRANSPARENT DEVELOPMENT PRICING
Pay for work that checks out.
No payment card. No live transfers. Try the full accounting flow locally.
TEXT EMBEDDINGS · SIMULATED$1.00 / 1,000 texts
Default 70% contributor share. Configurable pricing.
A few things to know.
Do I need a GPU to contribute?+
No. The first workload runs on CPU. GPU hardware may be detected, but GPU execution is not enabled.
Is this one giant supercomputer?+
No. The platform schedules independent inference batches across devices. It does not turn home networks into a tightly connected training cluster.
Can I earn real money now?+
All rewards and payouts in this version are simulated. Real billing and withdrawals are not connected, and income is never guaranteed.
Can I submit private company data?+
No. Only public and synthetic inputs are allowed. Home-PC owners can inspect workloads and data; confidential execution is not implemented.