TasteScore is InSoBlok's plug-in trust decision layer for smart wallets, protocols and AI agent payments. Before value moves, it brings wallet intelligence, counterparty relationships and transaction simulation into one explainable recommendation — ALLOW, REVIEW or BLOCK.
“Make sure AI agents don't send money to the wrong place.”
Where AI infrastructure gives agents the compute they need, InSoBlok owns the adjacent, more defensible lane: making sure those agents don't transact with the wrong counterparty.
| AI infrastructure | InSoBlok trust layer |
|---|---|
| AI infrastructure | Trust infrastructure |
| AI models | Wallet intelligence models (TasteScore V2) |
| GPU / compute | Blockchain data & analytics infrastructure |
| AI agents | AI agents + autonomous wallets |
| Model / API marketplace | Trust API marketplace |
| AI payments | Trust-aware transaction execution |
| Determines which AI resource to use | Determines whether a wallet/counterparty should be trusted |
Autonomous finance is more than compute. Every layer an AI agent touches — inference, identity, payments, and agent-to-agent settlement — runs through a live trust check on InSoBlok. It adopts the emerging agent standards (ERC-8004 identity, x402 payments) and wires trust into each one.
| Autonomous finance component | InSoBlok |
|---|---|
| AI inference | ✅Supported: Verifiable inferenceTasteScore V2 model outputs, on-chain anchored and stake-slashable |
| AI agents | ✅Supported: NativeTasteAgent plus an embeddable Trust Decision Engine any agent can call |
| Agent identity | ✅Supported: ERC-8004 compatibleOn-chain agent identity backed by the TasteScore reputation registry (AgentReputationRegistry: getReputation / getValidation / isAllowed) |
| Agent payments | ✅Supported: Trust-gated x402Every agent payment is trust-verified before it clears (x402 verify/settle: ALLOW→200, REVIEW/BLOCK→402) |
| Agent-to-agent economy | ✅Supported: SecuredAgent-to-agent settlement with a slashable on-chain reputation bond |
An AI agent wants to send 200 ETH (~$500K). Before executing, it asks TasteScore: EXECUTE, REVIEW or REJECT?
This is not a rebuild. It's an aggregation layer on top of four production systems, plus the on-chain oracle pattern already deployed for TasteScore.
TasteScore Trust sits between the agent and the blockchain. Every transaction is gated by a governance-configurable threshold before it ever executes.
The early phases deliver live, metered token utility. Later phases deepen the moat and open new revenue surfaces.
InSoBlok is building the trust layer for autonomous finance — the safety check between an agent's intent and an on-chain transaction.
AI Agent → Discover → Analyze → Trust → Execute → Monitor.
InSoBlok owns Trust.
Trust Infrastructure for AI Agents & Autonomous Finance