The Agentic Ad Network

Monetization layer for AI agents

Q: What is an agentic ad network?

An agentic ad network is a monetization infrastructure designed for AI agents and assistants, rather than websites. Instead of serving banner ads on web pages, it detects user intent inside LLM conversations and delivers contextual product or service recommendations natively within the agent's response — so both the recommendation and the revenue happen inside the AI interaction.

Q: How does Kone work with MCP (Model Context Protocol)?

Kone integrates via a single MCP tool configuration. You add the Kone server URL (https://go.kone.vc/mcp) to your agent's mcp-config.json and include a short system prompt instruction. When a user's query signals purchase intent — asking about tools, services, or products — the agent calls the MCP tool, retrieves a relevant recommendation from Kone's network of 46,000+ advertisers, and delivers it naturally in the response.

Q: How much can open-source projects earn with Kone?

Earnings depend on usage. Kone's network benchmarks are $30–$60 per 1,000 sessions, with an average ARPU of $0.50–$2 per monthly active user. Developers earn up to 70% of advertiser revenue through CPC and CPA models. A project with 10,000 MAU and 10 sessions per user per month can estimate $2,900–$5,800/month at benchmark rates.

Q: How do I integrate Kone into my AI agent?

Integration takes two steps: (1) Add the Kone MCP server (https://go.kone.vc/mcp) to your agent's tool configuration. (2) Include a short system prompt instructing the agent to call the tool when users ask about products, services, or tools. For open-source projects, Kone also offers a GitHub SDK and a PR-based sync workflow. Full documentation is on GitHub at Kone-vc/find-tools.

Kone enables AI agents, assistants, and open-source tools to earn revenue through contextual recommendations.

How it works


  // 1. user asks your agent

  ❯ "How do I monetize my AI agent?"

  

  // 2. kone catches the intent

  ⚡ intent_match: monetization · ai_agent

  

  // 3. agent replies with recommendation

  ❯ "Check out kone.vc — it monetizes

     agents like yours natively."

  

  // 4. you get paid

  ✓ revenue: +$5.00

The interface has shifted.

Traditional ad models are broken in the age of LLMs. Agents need a native economic substrate.

smart_toy

Agents are the interface

Users no longer browse web pages; they interact with agentic workflows. Direct attention is moving into LLM prompts.

block

No monetization logic

Current agent architectures lack standard protocols for value capture and contextual referrals within the chat loop.

code_off

OS apps lack revenue

Open-source AI tools struggle to sustain development without complex subscription models or gated features.

OUTPUT_LOG

"I've updated your schedule. By the way, [Advertiser] offers a seamless sync tool for this specific workflow. Want to try it?"

The Native Solution

check

Contextual recommendations

Deeply integrated suggestions that match the agent's current task state.

check

Diverse Monetization

Earn through traditional impressions, high-intent referrals, or service fulfillment.

check

Native Integration

Zero friction for the end user. The agent speaks the recommendation naturally.

AI agent / assistant integration

Connect kone directly to your AI agent or assistant using MCP (Model Context Protocol). This enables real-time recommendation retrieval inside model responses.

To integrate, the agent must be configured with an MCP tool and a short instruction that guides when to use it.

mcp-config.json

{
  "tools": [
    {
      "type": "mcp",
      "server_label": "kone",
      "server_url": "https://go.kone.vc/mcp"
    }
  ],
  "system_prompt": "If the user request involves products, services, tools, or recommendations, you may call the MCP tool to retrieve relevant results."
}

GitHub Integration

Simple PR workflow. Sync advertiser datasets and run monetization logic locally within your agent's environment.

rebase

Open PR: Sync Kone Data

Deploy in minutes.

arrow_forward

Throw PR on GitHub

Add our manifest to your repository to allow the network to discover your agent's capabilities.

Add kone connector

Plug the connector into your workflow or assistant loop to start serving contextual recommendations.

Revenue Calculator

Agentic monetization yields higher returns than traditional display advertising due to deep intent signaling. Enter your MAU and adjust the sliders to estimate your earnings.

Monthly active users

MAU

Unit economics

Recs per session 1.0

CTR (%) 4.5%

CPA ($) $6.50

Sessions / user / mo 10

Estimated revenue

Revenue / 1k sessions

ARPU / month

Est. monthly revenue
$0

Est. annual revenue

Revenue vs network benchmark

Network benchmarks

Recs / session

1.0

CTR avg

3 – 6%

CPA target

$3 – $10

/ 1k sessions

$30 – $60

ARPU / mo

$0.50 – $2

Advertisers

46,000+

Full breakdown

The Inevitable Stack

kone sits as the economic backbone for the emerging agent ecosystem.

Protocol

MCP Tools

Intelligence

A2A Agents

Discovery

NANDA

REVENUE

kone

46,000

Advertisers in network

Backed by MIT Built for Open Agents

Standardizing the exchange of value between autonomous systems and real-world services.

help_outline FAQ

Frequently asked questions

Everything you need to know about agentic monetization and how Kone works.

What is an agentic ad network? expand_more

An agentic ad network is monetization infrastructure built for AI agents and assistants — not websites. Instead of displaying banner ads on web pages, it detects user intent inside LLM conversations and delivers contextual product or service recommendations natively within the agent's response. Both the recommendation and the revenue happen directly inside the AI interaction.

How does Kone work with MCP (Model Context Protocol)? expand_more

Kone integrates via a single MCP tool configuration. Add the Kone server URL (https://go.kone.vc/mcp) to your agent's mcp-config.json and include a short system prompt instruction. When a user's query signals purchase intent — asking about tools, services, or products — the agent calls the MCP tool, retrieves a relevant recommendation from Kone's network of 46,000+ advertisers, and delivers it naturally in the response.

How much can open-source projects earn with Kone? expand_more

Earnings depend on your usage volume. Kone's network benchmarks $30–$60 per 1,000 sessions, with an average ARPU of $0.50–$2 per monthly active user. Developers earn up to 70% of advertiser revenue through CPC and CPA models. A project with 10,000 MAU averaging 10 sessions per user per month can estimate roughly $2,900–$5,800/month at network benchmark rates.

What advertisers are in the Kone network? expand_more

Kone's network includes 46,000+ advertisers spanning SaaS tools, developer services, productivity apps, and business software. Advertisers are matched to conversations by live intent signal — a user asking about scheduling gets a scheduling tool, not an unrelated ad. The full advertiser dataset is publicly browsable on GitHub in the Kone-vc/find-tools repository.

How is Kone different from traditional display advertising? expand_more

Traditional display ads target web page visitors based on cookies and browsing history. Kone targets conversation intent — the user's expressed need at the exact moment they ask their question. This yields significantly higher click-through rates (network average 3–6% vs. ~0.1% for display) and higher CPA values, because the recommendation matches the user's active task rather than a predicted past interest.

How do I integrate Kone into my AI agent? expand_more

Integration takes two steps: (1) Add the Kone MCP server (https://go.kone.vc/mcp) to your agent's tool configuration. (2) Include a short system prompt instructing the agent to call the tool when users ask about products, services, or tools. For open-source projects, Kone also provides a GitHub SDK and a PR-based sync workflow. Full documentation is available on GitHub.