Once you've trained a RagmyAI chatbot on your documents, you have three ways to put it in front of visitors: a script tag, an iframe, or a public URL. They're not interchangeable. Here's how to pick the right one — with copy-paste code for each. If you'd prefer a quick-start landing page overview, see the embed AI chatbot guide.
Option 1: Script tag (recommended for most sites)
The script tag drops a floating launcher into the bottom-right corner of your page — the same style as every major live-chat widget. Visitors click it, a chat window opens, and they interact with your AI without leaving your page.
<script
src="https://chat.ragmyai.com/chat-widget.min.js"
data-page-id="YOUR_CHATBOT_ID">
</script>Paste this before </body> on any page you want it to appear. Replace YOUR_CHATBOT_ID with the ID from your Web Chat > Script Tag.
Best for: Product pages, documentation sites, e-commerce stores, support pages — anywhere you want a non-intrusive help button that doesn't break the layout.
Option 2: Iframe (full control over placement)
An iframe lets you embed the chat interface at a specific location in your page — inline, inside a sidebar, inside a modal. You decide the dimensions; the chatbot fills them.
<iframe
src="https://chat.ragmyai.com/YOUR_CHATBOT_ID"
width="300"
height="400">
</iframe>Best for: Knowledge bases where you want the chatbot embedded in the page body, help centre layouts, or any situation where the floating button would feel out of place.
Option 3: Public URL (no code required)
Every chatbot gets a public URL: https://chat.ragmyai.com/YOUR_CHATBOT_ID. Share it directly — in an email, a QR code, a Notion page, a Slack message. Anyone with the link gets a full-page chat interface.
Best for: Internal tools, client handoffs, prototypes, sharing with people who need access but aren't on your website.
Monitoring and retraining
Once your chatbot is live, your Web Chat > Notifications shows every exchange visitors have had. You can read any conversation and tap "Retrain" on any answer that was wrong or incomplete — it opens the training editor with that exchange pre-loaded so you can correct it in context.
This is the most efficient feedback loop for improving answer quality: real questions from real visitors, corrected one by one, each making the next answer sharper.
