# 5. How It Works: The Agentic Layer In Detail
At the heart of Neox lies an **agentic coordination system,** a network of autonomous agents that operate continuously to analyse, forecast, and optimise yield performance across the global liquidity landscape.
These agents form Neox’s **intelligence layer**: a decentralised orchestration framework where data, risk, and liquidity are constantly monitored and adjusted to maintain security and efficiency at scale.
> **Every movement of capital within Neox is the result of data-informed, autonomous decision-making.**
***
### 5.1 The Three Agents
Neox’s agentic framework is powered by three core systems - the **Yield Agent**, the **Risk Agent**, and the **Routing Agent,** each performing a specialised function while coordinating in real time through shared data feeds and performance policies.
***
#### **🧠 Yield Agent: Forecasting and Strategy**
The Yield Agent is responsible for **predicting short-horizon yield movements** (up to 24 hours ahead) and proactively positioning liquidity to capture performance before market conditions shift.
It analyses a wide range of signals, including:
* Liquidity depth and utilisation ratios across major protocols
* Historical yield performance and volatility trends
* Capital inflows and outflows within on-chain and RWA markets
* Macro-level interest rate shifts and tokenised T-bill movements
* Cross-chain gas costs and latency metrics
By modelling these variables, the Yield Agent forecasts where yield is likely to rise or compress adjusting allocations to stay ahead of the curve.
> **It doesn’t chase yield, it anticipates it.**
***
#### **🛡️ Risk Agent: Constant Monitoring and Protection**
The Risk Agent operates as Neox’s **defence and compliance layer**. It evaluates every potential allocation through a multi-dimensional lens of protocol safety, liquidity strength, and counterparty risk.
Its live monitoring framework tracks:
* Stablecoin price integrity and de-peg thresholds
* Protocol audits, governance events, and security incidents
* Liquidity pool depth, utilisation spikes, and oracle accuracy
* Bridge uptime, latency, and reconciliation status
* Regulatory or issuer risk for tokenised RWA assets
If the Risk Agent detects an anomaly such as liquidity drain, volatility spike, or oracle deviation it automatically:
1. Freezes new allocations to the affected venue.
2. Triggers rebalancing through the Routing Agent.
3. Updates risk parameters network-wide.
In other words, it acts as Neox’s **circuit breaker** maintaining stability in volatile markets and protecting user capital within the parameters set by each client.
> **The Risk Agent ensures every yield decision is bound by policy, not speculation.**
***
#### **⚙️ Routing Agent: Real-Time Liquidity Movement**
The Routing Agent serves as the execution layer, the system that **moves liquidity across markets, chains, and pools** based on inputs from the Yield and Risk Agents.
Its responsibilities include:
* Real-time capital reallocation to the most efficient yield venues.
* Chain selection using interoperability protocols such as CCIP, CCTP, and USDT0.
* Transaction batching and gas optimisation for efficiency.
* Reconciliation of movements with policy and audit logs.
Because the Routing Agent is continuously connected to both on-chain and tokenised real-world markets, it can balance performance and safety dynamically ensuring capital is always deployed where it earns the best **risk-adjusted** returns.
> **Liquidity moves where logic dictates, not where humans guess.**
***
### 5.2 The Feedback Loop
These three agents operate in continuous dialogue, sharing data and recalibrating each other’s models in a **closed feedback loop**:
1. The **Yield Agent** forecasts changes in yield opportunity.
2. The **Risk Agent** assesses whether those opportunities fit current safety and policy parameters.
3. The **Routing Agent** executes reallocation if approved, and confirms the transaction’s effect back into the data layer.
This constant feedback process creates a **living system,** one that improves performance and resilience with every cycle.
***
### 5.3 Transparency Through Intelligence
Unlike opaque vault systems that hide allocation logic, Neox’s agentic framework is **observable by design**.
* Every agent action from allocation decisions to rebalancing events is recorded on-chain or in verifiable logs.
* Partners can view real-time exposure, performance history, and event-level data via dashboards or API.
* Predictive models are bounded by client policy meaning institutions retain control while benefiting from autonomous execution.
> **Transparency is not an afterthought, it is embedded in every decision Neox makes.**
***
### 5.4 Why Agentic Infrastructure Matters
In traditional systems, yield is passive.\
In Neox, yield is **active,** informed by data, guided by policy, and executed by code.
This agentic framework is what makes Neox distinct:
* It transforms yield from a static return into a programmable, autonomous flow.
* It scales safely across chains and markets without requiring manual oversight.
* It makes stablecoin yield infrastructure both **intelligent** and **trustworthy,** a first for financial institutions and neobanks seeking on-chain access.
> **Neox’s agents form the nervous system of the world’s next financial network, continuously learning, protecting, and optimising capital in real time.**
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://neox-3.gitbook.io/neox-docs/overview/5.-how-it-works-the-agentic-layer-in-detail.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.