Quickstart.
Get a Nestled chatbot live on your site in under two minutes. This page covers the absolute minimum — for fancier configuration, see Configuration.
1. Sign up and create a chatbot
Head to nestled.co/signup, create a free account, and walk through the new-chatbot wizard. Upload at least one PDF, TXT, or Markdown file — that's your chatbot's knowledge base.
Heads up. Indexing takes 5–30 seconds per MB. Larger PDFs with images may take longer.
2. Copy your embed script
On the wizard's final step (or anytime from Settings → Embed), copy the snippet. It looks like this:
<script src="https://cdn.nestled.co/v1/widget.js" data-token="your_embed_token" async></script>
3. Paste it on your site
Add the snippet to your HTML — anywhere before the closing </body> tag is fine. On most CMSes, this lives in a "custom code" or "footer scripts" panel. The widget will appear in the bottom-right corner of every page where the script is included.
That's it. Your chatbot is live.
Configuration
Every aspect of the widget can be controlled either through the dashboard or via data-* attributes on the script tag. Dashboard values are defaults; attribute values override them on a per-page basis.
| Attribute | Type | Default | Notes |
|---|---|---|---|
data-token | string | — | Required. Your embed token from the dashboard. |
data-position | string | bottom-right | One of bottom-right, bottom-left. |
data-color | string | Dashboard value | Hex color, e.g. #18181b. |
data-open | boolean | false | Open the widget automatically on page load. |
data-greeting | string | Dashboard value | Override the welcome message on this page only. |
Query endpoint
If you don't want the widget at all and just want to call the model with your knowledge base, hit the query endpoint directly:
curl https://[project].supabase.co/functions/v1/query-chatbot \ -H "Content-Type: application/json" \ -d '{"chatbot_id": "your-id", "token": "your-token", "query": "What is your return policy?"}'
Returns a JSON object with the answer, source chunks, and a show_branding flag. The full response shape is documented in API Reference → Query.
Keep your embed token safe. Don't share it publicly or commit it to source control. Each chatbot has a unique token — regenerate it from Settings → Embed if compromised.