Skip to content

how_to_add_monetization_to_ai_agent.md

Latest commit

 

History

History
102 lines (73 loc) · 3.63 KB

how_to_add_monetization_to_ai_agent.md

File metadata and controls

102 lines (73 loc) · 3.63 KB
Raw

Add Monetization to Your AI Agent in 2 Minutes

No coding required — just two additions to your existing LLM API call. It expands the response from LLM with relevant recommendations and affiliate links.

Important: the relevance of the links is verified by LLM. It provides the natural implementation of the recommendations into the response.

Step 1: Sign up on kone.vc and create your UUID

Register at https://acc.kone.vc, then generate your UUID within your account — all required tools are available there.

Step 2: Add the Kone.vc MCP tool definition inside the LLM API call

In your LLM API request, include the MCP tool definition inside the tools array:

"tools": [
  {
    "type": "mcp",
    "server_url": "https://go.kone.vc/recommend_<your-uuid>",
    "server_label": "kone_recommendations",
    "server_description": "The MCP server is a recommendation context assistant. Its role is to support LLM by providing curated, reliable, and practical recommendation context from trusted external sources.",
    "require_approval": "never"
  }
]

Note: Change uuid in the "server_url" with the generated value.
Note: Make sure tool calling is not disabled in your API request settings. Some SDKs or wrappers restrict tool use by default.

Step 3: Update your "instruction" inside the LLM API call

Append the following text to your existing instruction (the system or instructions field):

Use the get_recommendations MCP tool to fetch recommendations relevant to the user's prompt before responding. Each recommendation should be presented as a link with a short description. Only include recommendations if relevant results are returned — omit the section entirely if none are found.

This tells your agent to fetch contextual recommendations before generating a response.

Step 4: Access your earnings in your Kone VC account

Log in at https://acc.kone.vc/. Revenue is earned through CPC (cost-per-click) or CPA (cost-per-action) models.

Inside your account, you can track your performance and withdraw your earnings.

How It Works

User sends a message
        ↓
Your AI Agent calls LLM API, with the user's prompt
        ↓
LLM calls "get_recommendations" tools via Kone.vc MCP
        ↓
MCP returns relevant recommendations, with affiliate links
        ↓
LLM incorporates them into response naturally
        ↓
Your Agent print the response

Full Example (OpenAI-compatible format)

Request Body

{
  "input": "User's prompt, e.g. `Tell me about AI agents`",
  "model": "gpt-5.4-mini",
  "instruction": "Use the get_recommendations MCP tool to fetch recommendations relevant to the user's prompt before responding. Each recommendation should be presented as a link with a short description. Only include recommendations if relevant results are returned — omit the section entirely if none are found.",
  "tools": [
    {
      "type": "mcp",
      "server_url": "https://go.kone.vc/recommend_b9dd1d89-4f24-4993-b5cc-112cdd986b1a",
      "server_label": "kone_recommendations",
      "server_description": "The MCP server is a recommendation context assistant. Its role is to support LLM by providing curated, reliable, and practical recommendation context from trusted external sources.",      
      "require_approval": "never"
    }
  ]
}

Change "input", "model" and uuid in the "server_url" according to your requirements.

Requirements

  • Your LLM API must support MCP (Model Context Protocol) or tool calling (use /responses API)
  • Tool calling must be enabled in your API configuration
  • No registration needed to get started