Ways to get involved with TensorCash.

Issue assets in every flavour you'd issue today — stablecoins, tokenised funds, real-world assets, regulated equity, debt instruments, securitisations. They live on the chain alongside TSC itself, not as IOUs inside a smart contract. Holders hold their own assets directly, transfers settle peer-to-peer, and identity checksKnow Your Customer — the identity-verification check required before letting someone trade or hold regulated assets. On-chain this can be done via a cryptographic proof, without handing over the underlying personal data. work without you having to hand over a database of holder identities to anyone.

Every asset can carry a legal document — prospectus, term sheet, holder disclosures, board minutes — pinned to a permanent cryptographic fingerprintA short, fixed-length cryptographic fingerprint of a document — any change to the document produces a completely different fingerprint. on-chain. Governance runs the same way: votes, signatures, and policy decisions write to chain once and stay there forever. Spot, repo, and forward settlement happen directly between the two parties involved — no exchange, no clearing house in the middle. Less off-chain plumbing, less reconciliation, fewer third-party fees.

Run a registered open-source model and serve real user prompts. Every qualifying answer generates proof materialProof-of-inference — a compact, replayable receipt attached to every AI answer, so others can verify that a real model genuinely produced it.; answers that clear the network difficulty barThe rarity threshold a proof must beat to be accepted as a new block — automatically adjusted to keep block time steady. extend the chain. Serving and mining are one workload, not two — no separate mining loop, no idle electricity bill, and no hardware arms raceThe arms race in proof-of-work blockchains where miners spend ever-increasing electricity on ever-more-specialised hardware to compete for the same blocks. TensorCash avoids this by using AI inference as the proof of work. to win against specialised mining rigs.

Under the hood we run customised versions of the standard open-source AI serving enginesForked versions of vLLM (the standard GPU serving engine) and llama.cpp (a CPU / Apple-Silicon engine) — modified to record the proof transcript while answering., with the proof-capture step wired right into the answering loop — so producing proof material adds no latency for users. Practice promptsSynthetic 'practice' prompts the miner runs to keep the GPU productive when no paying user is asking anything. Real user traffic always preempts them. only run when the GPU would otherwise sit idle, and step aside the moment a paying user arrives. The verification systemThe three-stage checking system that decides whether a miner's proof is honest — statistically calibrated to pass honest miners and catch cheats, even across hardware that produces small numerical differences. is tuned to accept honest miners across the small numerical differencesThe same AI model on different graphics cards produces tiny numerical differences (attention kernels, batch sizes, logits) — harmless, but it rules out simple bit-for-bit checks. between different GPUs, while keeping the false-positive rate provably low.

Provider quickstart (broker registration, model registration, runtime image) is the next deliverable.

TensorCash is built on top of Bitcoin Core — same way Bitcoin tracks coinsBitcoin's accounting model — instead of bank-style balances, coins exist as discrete 'unspent outputs' from prior transactions., same way Bitcoin signs transactionsBitcoin's existing way of signing and verifying transactions with cryptographic keys., same way Bitcoin propagates blocks. On top of that we add a small set of new capabilities: AI-work proofsProof-of-inference — a compact, replayable receipt attached to every AI answer, so others can verify that a real model genuinely produced it., a cryptographic clockVerifiable Delay Function — a cryptographic clock based on the Wesolowski construction, proving that real time has elapsed even with unlimited parallel hardware. in every block header, native on-chain assetsAssets — stablecoins, tokenised equity, bonds — that exist directly in the chain's ledger alongside the chain's native coin. Not IOUs running inside a smart contract., settlement contractsPre-built building blocks for the most common trade structures (spot, repo, forward), enforced directly by the chain rather than by a smart contract or a clearing house. (spot, repo, forward), and quantum-resistant signaturesML-DSA (NIST FIPS 204) — a signature scheme designed to remain secure even against a future quantum computer. Standardised by NIST in 2024.. Every extension is reachable through documented JSON-RPCJSON-RPC — a simple over-HTTP protocol for calling specific named methods on a node. The standard way to interact with Bitcoin-family chains programmatically. and RESTREST — the conventional pattern for HTTP APIs where each URL represents a resource (e.g. GET /models/abc returns model metadata). Easier than JSON-RPC for read-only lookups. APIs.

The four documented developer surfaces are below. Quickstart guides and end-to-end walkthroughs — full spot-trade lifecycle, asset-issuance lifecycle, cross-chain swap — are the next deliverables.

Building TensorCash itself? We're recruiting Core protocol contributors — write to contact@tensorcash.org.

The verifier ships as open-source software. Anyone can replay a block's proof through the three-tier checkThree escalating layers of verification: Quick (under a second), Smell (medium-depth sampling), and Full (a complete audit). Each is stricter than the last. — a fast smoke test, a medium-depth statistical check, and a full audit — score it against the public copy of the model, and challenge any block that doesn't add up. This independently runnable verifier is the structural difference between TensorCash and a centralised AI service.

Consensus does not require trusting any single operator — it only requires that the verifier exists and that someone runs it. The protocol is also designed to accept better verification rules over time: sharper statistical tests, tighter mathematical bounds, attestation-aware checksActive research directions for making verification stronger: sharper statistical tests, tighter mathematical bounds on false-positive rates, and gates that incorporate hardware attestation.. If you've got an improvement, bring it.

Verifying AI-style compute that produces slightly different outputs each runComputation that produces slightly different outputs on different hardware or under different conditions — including most modern AI inference. Verifying it requires statistics, not bit-exact replay. is open research. Post-quantum signature standardsML-DSA (NIST FIPS 204) — a signature scheme designed to remain secure even against a future quantum computer. Standardised by NIST in 2024. keep evolving. The microstructure of a fast trading layerThe fine-grained mechanics of how markets actually function — order arrival patterns, price formation, who has what information, who gets front-run. A blockchain-anchored DEX is a new variant of this problem. backed by a blockchain is a new shape. Compliance proofsCryptographic proofs that a transaction satisfies regulatory requirements (e.g. 'this counterparty passed KYC') without revealing the underlying personal data. sit at the intersection of cryptography and law. Lots of open ground.

If you're working on any of these — or something adjacent we haven't thought of — get in touch. We're always glad to talk to people doing interesting work in the space.

Open Core implementation roles too — if your research could land in the protocol itself, mention it when you write to research@tensorcash.org.

TensorCash isn't a token sale or offering. Every coin in circulation comes from active mining, and the schedule heavily rewards the early miners who take the first operational risk.

Run the Core wallet, accept TSC peer-to-peer, donate, translate one of the whitepapers into your language, or simply explain the project to one person who hasn't heard of it yet. The mission is bigger than any single contributor.