August 20, 2026
With smart contracts increasingly dependent on AI, Truebit provides proven integration that you can trust.
Verified Agents
Verification for offchain agents
Transform black-box AI calls into a chain of provable steps
Verifiable AI
Verifiable and accountable AI
A few years ago, an offchain dependency mostly meant a price feed. Today it increasingly means a call to an AI model, or a multi-step workflow spanning several APIs — a judgment call too complex for a static oracle, but one that still has to be provable.
Truebit transparently executes critical workflows. Each step – the trigger, the orchestration logic, the AI model and API calls, and the deterministic logic that determines the result — gets its own cryptographic transcript, chained into one end-to-end auditable record.
Trustless Agents
Proof for actions you Can’t take back
When AI decisions settle a market, release funds, decide who’s right, or move capital, the stakes are too high to blindly trust an answer. You need proof. Truebit turns each of those decisions into a transcript: what was checked, by what rule, before anything moved.
Settlement
When an AI model determines an outcome — who won a market, whether a claim was valid — that determination has to be provably independent, not just correct. Truebit settles AI-adjudicated prediction markets without asking traders to trust a black box.
Escrow & payments
Release funds only when a verified condition is actually met — a delivery confirmed, a milestone hit, a threshold crossed — instead of trusting whichever party controls the wallet. Every release comes with a transcript of exactly what was checked.
Adjudication
When two parties disagree about what happened, an AI can weigh the evidence — but the ruling only holds if neither side could have shaped it. Truebit runs the evaluation independently and produces a record either party can check.
Agentic trading
An autonomous agent executing trades needs bounded, provable limits — not just good intentions. Every decision, and the logic that authorized it, is captured in a transcript before capital moves.
Proof of AI performance
Verifiable execution
Every AI model call runs inside a Trusted Execution Environment and is independently auditable — proof that inference executed exactly as intended, with nothing altered between the call and the result.
Auditable decision trails
Every prompt sent and every response returned is captured in a cryptographic transcript — a durable, checkable record of exactly what the model was asked and what it said back.
Data provenance & integrity
Trace exactly which model or API was called, with what parameters, and when — provenance for every offchain dependency your smart contract relies on, not just the AI call.
Cross-chain & API integration
Seamlessly access AI models through Dynamic Oracles, with orchestrated calls to private APIs and other Web2 systems, then settle the result on any blockchain — one workflow, one chain of transcripts.
Security you can prove
Every check verified, every assumption challengeable, every result immutable. Eliminate silent failures, manual errors, and audit guesswork.
Future-proof your model
Swap the model, add a new API, or change the decision logic without redeploying your smart contract — every piece of the pipeline is independently versioned and upgradable.
Unlock verifiable, trustworthy agents
Move beyond “just trust us” to “just audit us” for every AI call your smart contract depends on.