Website-trained chat
Turn your website into a chatbot.
Kanha crawls your pages, trains a model on your content, and runs it on the visitor's device.
Product
From website content to an embedded bot.
A single workflow handles crawling, dataset preparation, model training, testing, and installation.
Crawl to dataset
Keep each stage clear.
Select the source pages, prepare the dataset, and start training from the same project.
Site project
Crawl
Choose the pages that contain the answers.
Prepare
Turn page content into a training dataset.
Train
Create a model tied to your selected content.
Install with the interface that fits your site.
Use the script embed, Web Component, or React package. The bot configuration stays in Kanha while the chat surface lives on your site.
<script
type="module"
src="https://cdn.jsdelivr.net/npm/kanha-ai/dist/widget.js"
></script>
<kanha-bot model-url="YOUR_MODEL_URL"></kanha-bot>How it works
A training job you can follow.
The dashboard exposes each stage instead of hiding the work behind a single loading screen.
New bot workflow
Status updates by stage
- 01
Crawl
Choose a site and the pages that should become source material.
Pages indexed - 02
Prepare
Kanha turns the indexed content into a training dataset.
Dataset ready - 03
Train
Start a training job and follow its state from the dashboard.
Model created - 04
Embed
Test the result, copy an embed key, and add chat to your site.
Ready to install
The chatbot runs on the visitor's device.
Kanha packages the trained model for the runtimes you support. The chat can answer locally without sending each question through a hosted inference API.
Build a test bot from your own site.
Create an account, add a site, and choose the pages to crawl. The free flow prepares the content and starts a training job you can follow from the dashboard.