Infrastructure for the consented collection and monetization of behavioral data —
insights as a service, to businesses and to users.
People opt in to share behavioral signals from their phones and are credited for what they share; from token launch, they share in the revenue. Businesses buy insights, never the data: there is no raw-data export endpoint, by construction.
Important
Status — November 2026. Pre-revenue. No users yet. No outside funding. Nobody has paid yet. The Telegram Mini App is in development. Much of what follows is design — the text says which phase each item belongs to.
One infrastructure, three sides:
| Who | What they get | Phase | |
|---|---|---|---|
| 1 | People | Opt in, choose what to share, get credited for the signals they share. From token launch, they share in the revenue. Insights about their own patterns arrive in Phase 2. | 1 → 2 |
| 2 | Businesses | Answers to their questions as a service — never raw data, never individual profiles. | 1 → 2 |
| 3 | AI teams | Federated learning, without transferring data. | 3 |
The Phase 2 design for businesses: a business asks a question in plain language, and AI finds links across anonymized behavioral vectors and shows what is likely to happen next in a similar situation.
Phase 1 starts simpler: aggregated reports on activity rhythm, timing and transitions between categories.
Note
What it is not. Not a data marketplace — there is no export endpoint. Not a token-emission scheme — the token has no mint function, and the referral is one level, 10%.
Businesses today buy datasets and logs of past behavior, then pay analysts to decode them. ForeMetric is designed to reverse this chain:
Business asks a question in plain language
↓
System searches anonymized behavioral vectors from people who opted in
↓
Finds links between events that look unrelated
↓
Returns an insight with a recommendation:
what is likely to happen next in a similar situation
The business buys the answer — not the data, not the people. The people whose signals formed the answer share in the revenue.
Warning
This is Phase 2 design. It is not built, and its predictive accuracy has not been measured. Phase 1 is the cheap test of the conditions it depends on: people opt in, people come back, a business pays.
The narrowest wedge. Phase 1 starts with a Telegram Mini App as an onboarding funnel — distribution, consent, a TON wallet, one-level referrals — and tests, cheaply, three things:
- Do people opt in?
- Do they come back?
- Does a business pay?
Enrollment is open. One cohort, no separate pools and no invite curation: a user is a user, whatever brought them here. Filtering happens at the data-weight layer, not at the door.
The first paid pilot is an aggregated engagement report (cohort level, prepared by the team) for consumer-fintech teams that buy traffic inside Telegram, priced at $1–5K.
This is a test of willingness to pay for an answer — not the product, and not an operating business model.
| Phase 1 — now | Phase 2 — designed, not started | Phase 3 — designed, no date | |
|---|---|---|---|
| Scope | Telegram Mini App as an onboarding and referral funnel: distribution, TON wallet, per-category consent, one-level referrals, IOU balance. Two signal categories only: Mini App session events and, optionally, DNS categories set up manually by the user | Native app with more signals collected on the device; attested natural-language queries for businesses; insights for users; token launch | Federated learning for AI teams; TON Storage as the only storage layer |
| Trust model | Disclosed trusted operator: we hold the encryption key (AWS KMS, with an audit trail of every access) and can in principle read the data | Hardware-attested computing, differential privacy, a hard minimum of 50 users per group | Decentralized attestation |
| Reports | Made by the team from PostgreSQL aggregations; the buyer receives an aggregate, never raw records or individual profiles | Attested natural-language queries answered with predictive insights | Same as Phase 2, with attestation verified by a quorum rather than a vendor |
Important
Phases advance on metrics, not on dates. The Phase 1 gate includes an open-enrollment alpha that measures what brings people back, a first paid pilot, and a secured runway. If a gate is not passed, the phase continues.
| Component | State | |
|---|---|---|
| 🟢 | Public prototype at foremetric.ai | Live — simulated data, marked as such |
| 🟢 | Identity Vault contract on TON Testnet | Live |
| 🟡 | Telegram Mini App | In development |
| 🟡 | Rust backend with PostgreSQL IOU ledger | In development — Mini App being connected to it |
| 🟡 | DoH/DoT DNS server | In development |
| ⚪ | Native app | Designed, not built |
| ⚪ | Confidential computing | Designed, not built |
| ⚪ | Predictive answer engine | Designed, not built |
| ⚪ | Revenue splitter | Designed, not built |
| ⚪ | Token launch | Designed, not built |
| ⚪ | Signal Passport · Oracle · Missions | Designed, not built |
| ⚪ | Federated learning | Designed, not built |
Warning
A specification only. No token contract or splitter is deployed. No income figure is promised. There is no token sale, presale or giveaway.
| Property | Value |
|---|---|
| Total supply | 1,000,000,000 — fixed |
| Mint function | None |
| Settlement | Immutable splitter contract on TON |
| Pre-launch balances | IOU entries in a ledger, converted 1:1 at launch |
| Launch | Coincides with Phase 2 — no committed date |
| P2P transfer burn | 0.1% — splitter distributions are exempt |
| Destination | Share | Note |
|---|---|---|
| Contributors | 80% | Of gross B2B receipts |
| Referrer | 10% | One level only |
| Treasury | 10% | 20% when there is no referrer |
The public front page of ForeMetric. Development happens in private repositories.
Caution
Earlier public experiments are superseded — do not deploy or use them.
This includes a Jetton prototype with an admin-controlled mint operation (OP_MINT). The current design has no mint function. Those experiments were never audited. No token contract is deployed.
Code in this repository is released under the MIT License — see LICENSE.