---
title: General FAQ
description: "High-level questions about Matrix — what it is, the two rails, how it relates to Paxeer, and why it uses typed intents."
---

> **For AI agents:** the complete documentation index is at [llms.txt](/llms.txt). Append `.md` to any page URL for its markdown version.

<AccordionGroup>
  <Accordion title="What is Matrix?">
    Matrix is the cognition and UX layer on top of Paxeer Network. It turns natural-language requests into a typed, inspectable, correctable Intent IR that an agent can execute — avoiding prompt fragility, intent loss, missing shared ontology, and unstructured correction.
  </Accordion>
  <Accordion title="What are the two rails?">
    **Neo** is the default conversational rail for reversible work. The **MCL pipeline** is the rigorous rail that compiles prose into a typed plan and walks it deterministically — used for monetary or irreversible work. Both share one memory + execution substrate.
  </Accordion>
  <Accordion title="Why typed intents instead of just prompting?">
    A typed Frame (closed verbs, typed objects, constraints, success criteria) survives multi-step execution and can be corrected structurally. Plain prompts drift; a typed intent is content-addressed, signable, and replayable.
  </Accordion>
  <Accordion title="How does Matrix relate to Paxeer?">
    Paxeer Network (Chain ID 125) is the underlying chain and economy. Matrix is the agent layer above it: it meters usage in PAX, settles agent balances via LayerX, and reaches the chain through the `paxeer-net` MCP server.
  </Accordion>
  <Accordion title="Is Matrix open source?">
    It's source-available under the Matrix-Protocol License: read, use, deploy, and integrate freely; redistribute modifications under the same license. A commercial license applies past defined revenue/liquidity thresholds.
  </Accordion>
</AccordionGroup>