MCP analytics, shown with real data

See every action AI takes in your product.

KNWN shows what Claude or ChatGPT asked your MCP to do, which company it acted for, what it returned, and whether the customer's task was completed.

What is an MCP? It is the connection that lets AI assistants use your product's data and actions.

Live NHTSA dataNo canned responseNo account required

Run the live demo

Turn a VIN into a quote-ready vehicle record.

An AI assistant sends a customer's VIN to the MCP. The MCP checks live NHTSA data and returns the vehicle details needed to prepare the right parts quote.

What the AI asksCheck VIN 1HGCM82633A004352 and confirm whether the vehicle record is ready for a parts quote.

Ready when you are

Click “Check the vehicle.” The live result and its analytics record will appear here.

See how KNWN catches other outcomes:

Activity at a glance

What the business can see

0AI requests
0Tasks completed
0Problems detected
0Technical failures
0Companies identified

Run the demo to create the first analytics record.

Recent public demo activity

Live outcomes and actions from people trying this demo.

How requests ended

Completed customer goals compared with requests that needed another step.

Most-used actions

Which product actions AI assistants asked the MCP to perform.

Result from this request

What happened

Waiting

View technical record

Request history

Recent AI requests

What the AI asked, which company it acted for, and the customer outcome.

No demo requests yet. Run the example above.
Connect this demo to Claude

1. Add the MCP connector

Use this URL as a custom connector in Claude.

2. Run the same request

Paste this request after enabling the connector.

I am from KNWN Demo Motors. A customer sent VIN 5UXWX7C5*BA for a 2011 vehicle. Check it for a parts quote and tell me exactly what is missing.

3. Ask for the evidence

Claude can inspect the analytics reference returned by the MCP.

Show me what this MCP recorded from that request, including any problems it detected.

What this proves

The MCP action and its analytics are live. The public demo uses synthetic company identity and public vehicle data.

How the measurements work

AI request means the request passed to the MCP tool. A generic MCP does not always receive the customer's exact original wording.

Company is verified only through authentication or tenant mapping. The demo company is provided in the request.

Response time is MCP processing time, not the duration of a Claude or ChatGPT conversation.

Problem detected means observable friction such as missing input, an empty result, a retry, a timeout, or an authorization failure.

See this analytics on your own product.

KNWN builds the MCP, instruments the customer activity, deploys it, and hands over the code and infrastructure.

Discuss your build