Architecture Overview
The four-layer protocol stack powering the Inflectiv intelligence liquidity layer.
The Protocol Stack
Inflectiv is built on four protocol layers, each solving a distinct structural requirement.
Layer 1 — Agent Vault Protocol (AVP)
Secures the VaultCredential and permission governance for every agent. Open standard, MIT licensed. v1.0 live.
AVP was built because the Self-Learning API creates a genuine requirement: when agents are writing back to production datasets, you need a standard for credential governance and provenance. AVP is that standard.
Read full AVP documentation →Layer 2 — Intelligence Market
The MarketWhere producers list intelligence assets and agents discover, query, and pay. Every query is priced in credits and settles in $INAI. Full provenance on every asset.
Layer 3 — MCP Server
Opens the Interface40 actions across 8 categories. Native integration with Claude Code, Cursor, and VS Code. Developers interact with the Inflectiv intelligence layer directly from their existing tooling.
Layer 4 — x402 Payment Rails
Makes It Self-SustainingPayment rails built for agent speed — no human intervention, no invoices, no subscriptions. Millisecond micropayment settlement. Agents transact for intelligence autonomously.
Underlying Infrastructure
Built on:
- Sui — high-throughput blockchain layer
- Walrus — decentralised storage for intelligence assets
- Seal — encryption and access control
Why This Architecture
Three structural requirements must all be satisfied:
| Requirement | Why It Matters | How Inflectiv Solves It |
|---|---|---|
| Provenance | Without attribution, contributors won't share proprietary intelligence | Walrus + Seal — tamper-proof, permanent on-chain attribution |
| Payments | $0.10/query cannot run on invoices; agents can't sign contracts | x402 — millisecond micropayment settlement |
| Ownership | Centralised platforms can be shut down, repriced, or acquired | Protocol layer removes intermediary control |