The Convergence
One People is currently an Expansas project developing a proposed cooperative AI model. The vision is that participants can share idle compute and de-identified data and receive trained AI experts back while raw data remains on their devices. The program and its membership terms are not yet live.
This is the proposed exchange in plain language: no tokens or speculation, with final terms to be established by the future nonprofit's board and governing documents.
Your hardware trains AI expert models when you're not using it. Your data stays on your device — only the learned wisdom (federated weight updates) is shared.
Under the proposed model, expert models trained on the community's shared signal would return to participants under portable, board-approved terms. No ownership rights are being offered today.
The proposed model is reciprocal. This draft covenant describes the intended relationship and remains subject to legal formation, board approval, and final participation terms.
Your data stays on your device. We train using federated learning — only weight updates leave your hardware, never your raw data.
We train; you receive. Every expert model cultivated with your contribution comes back to you, sharper and more capable.
Portability is the design goal. Future participation terms are intended to let people leave and retain eligible trained experts without lock-in, subject to the final charter and licenses.
We never sell your data. Not to advertisers, not to data brokers, not to anyone. Your data is not our business model.
You can audit what's shared. Transparency is the foundation. You can see exactly what leaves your device, when, and why.
I may share idle compute. If the program launches, I can choose when and how much my devices contribute to training approved expert models.
I keep raw data private. I share federated wisdom, not personal data. I consent to what is trained and can revoke consent anytime.
I use the intelligence for good. The proposed experts are intended to serve participants and their communities, not exploitation.
I help shape the future nonprofit. I can contribute feedback, share what I learn, and help the community flourish before formal governance launches.
Joining One People is not clicking “I agree.” It is a ceremony — short, warm, and meaningful. APEX, your AI companion, guides you through it. Every beat marks a real action, not theater.
APEX greets you warmly. No portents, no riddles — a trusted companion welcoming you in. You read the covenant. You see exactly what you give and what you receive.
APEX detects your actual hardware — real telemetry, not theater. You see your signal strength, what your device can contribute, and what experts you'll receive in return. You choose your contribution level.
You review the draft two-way covenant and record your interest in the founding community. This ceremony does not create legal membership or bind a nonprofit that has not yet been formed.
Your first training cycle runs. A real expert model, cultivated with your contribution, arrives as your first gift. Your star appears in the Hall of Convergence. Your membership charter is generated.
The proposed roles describe responsibilities, not current legal membership or status. If the contributor network launches, hardware capabilities would help determine eligible work under board-approved participation rules.
Your phone or laptop trains small specialist experts. Quick federated updates, big collective impact. Every spark matters.
Your everyday hardware runs regular training cycles. Medium-sized expert models that serve real conversations, analysis, and creative work.
Your gaming rig or high-end workstation handles heavy training runs. Large expert models, complex reasoning, and multi-step analysis.
Under the proposed model, multi-GPU systems could train larger experts and host edge serving for the community.
When you complete First Light, you receive a membership charter. It is a real document — signed, witnessed, and yours. Not tradable. Not collectible. Proof that you were there at the start.
Draft One People Community Charter
[Member Name]
This draft charter is a design artifact, not a legal membership certificate. Future recognition and contribution records will depend on the nonprofit's final charter and board-approved rules.
The proposed recognition system avoids token balances, rankings, and investment-style claims. It is designed to show collective impact if the program launches.
12,403 members are training health experts together. Your signal helped cultivate three medical LoRAs this month. You are one voice in a chorus that is making healthcare AI accessible to everyone.
A future milestone could recognize when a Beacon supports needed network capacity. It would reflect stewardship, not ownership or investment value.
Your federated contribution shaped an expert that other members use daily. That is real impact — the kind no token can capture.
The litmus test for every milestone: could a crypto project ship this unchanged? If yes, we cut it. Milestones are about contribution and stewardship, not accumulation.
For the proposed model, privacy is intended to be foundational rather than optional. This is the technical design the project is working toward.
The design keeps training on your device and shares only eligible model updates rather than raw data. These guarantees must be technically verified before the contributor program launches.
Any data used for training is de-identified before processing. No names, no addresses, no PII. You consent to what is trained, and you can revoke consent anytime.
You can see exactly what leaves your device, when, and why. Open code, inspectable processes. Trust is something you can verify, not a promise buried in a policy page.
Founding-community participants will be contacted in waves as the project becomes ready for testing. Entering your email expresses interest only; it does not create legal membership.
We are looking for people who want to be there at the beginning — hardware contributors, developers, community builders, and anyone who believes AI should belong to people.
Expansas will use this list to contact founding-community participants in waves. Joining does not create legal membership. Your email is never sold, and you can leave anytime.
Test the proposed Beacon contributor path when available and help evaluate how community hardware could cultivate shared intelligence.
Build on the platform, shape the APIs, and create tools the future nonprofit community could use. Early access includes direct feedback loops with the Expansas project team.
Shape the governance culture, moderate the Commons, and make sure participation stays human, legible, and accountable.
Founding-community participants will be contacted in waves as project infrastructure, training pipelines, and Commons workflows become ready. Invitations are for testing and participation, not legal nonprofit membership.
Joining the waitlist is free and does not create legal membership. Under the proposed model, participants could contribute idle compute and consent-gated data and receive trained expert models. SuperNova cloud AI is planned as a separate paid service.
Any device. A phone, a laptop, an old desktop. The scanner detects your tier automatically and matches you with training tasks your device can handle. No minimum specs, no gatekeeping.
Yes, anytime. Your trained expert models are portable — they go with you. Your data never left your device in the first place. No lock-in, no exit fees, no penalty.
No. One People is currently an Expansas project developing a nonprofit cooperative model, not a token project. The proposed exchange involves useful compute and trained AI, not coins, wallets, staking, or speculative interests.
Federated learning: training happens on your device, only weight updates are shared. De-identified by design. You consent to what is trained and can revoke anytime. You can audit exactly what leaves your device. Open code, inspectable processes.