> ## Documentation Index
> Fetch the complete documentation index at: https://docs.prado.network/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Agents (Finance)

### **Example: AI trading agent deciding BUY / SELL**

An AI agent monitors markets and decides whether to BUY, SELL, or HOLD an asset.

### **Flow**

1. The agent fetches external data

   (prices, indicators, market data APIs)
2. The agent sends a prompt to an AI model

   (e.g. “Given this data, should I BUY or SELL?”)
3. The AI model returns a response

   (e.g. BUY with a confidence score)
4. The agent executes the trade

### **Where Prado is used**

* **HTTP runtime**
  * Records the external data APIs the agent called
  * Records the AI model request and response (or hashes of them)
* **Optional JS / Python runtime**
  * Records deterministic post-processing logic
  * Example: reject low confidence, cap position size

### **What Prado proves**

* What market data the agent observed
* What prompt was sent to the AI model (or its hash)
* What the AI model returned
* Whether deterministic logic modified the result
* That this evidence was not altered later

### **What Prado does NOT prove**

* That the trade was correct
* That the model reasoning was sound
* That the outcome was profitable

### **Why this matters**

When AI agents make financial decisions, failures are inevitable.

Prado allows teams to **inspect and audit decisions after the fact**, instead of relying on logs or explanations.