Dendron tracks behavioral signals, scores visitor intent, and surfaces AI-generated Smart Cards at the right moment. 13KB. Zero server. Your LLM.
You scrolled, clicked, and read — Dendron tracked all three signals, scored your intent, and triggered this card. On a real site, your LLM would generate this content based on the visitor's behavior.
Try it on your site →Dendrites receive signals. The soma processes them. The axon fires Smart Cards. All in 13KB.
Scroll, reading time, clicks, cursor patterns. Six passive listeners, zero main-thread impact.
Context Engine scores engagement across 3 dimensions in a Web Worker.
When intent crosses threshold, Dendron calls your backend. Your LLM generates the card.
Smart Card appears in a Shadow DOM widget. Right message, right moment.
Pick your platform, choose your LLM, set one secret. That's it.
Try Dendron instantly with our hosted backend. No setup needed.
What you get:
One script tag:
Pick where to deploy your Dendron backend proxy. All have free tiers.
100K req/day free
Global edge. Fastest.
100K invocations/month free
Serverless. Auto-scales.
125K invocations/month free
Functions v2. Streaming.
Loading platform guide...
Drag a provider to the right. Or click to select.



Drag a provider here
or click to selectCopy the commands below to deploy your Dendron backend.
Two steps. Any framework. Under 3 minutes.
npm or CDN. Zero dependencies.
One callback — onQuery routes to your backend.
Follow the Quick Start above to deploy. Here's what each platform offers.
100K requests/day free
Global edge network
~10ms latency worldwide
Best for: Global audience
100K invocations/month free
Serverless functions
Auto-scales automatically
Best for: Next.js sites
125K invocations/month free
Functions v2 with streaming
Edge compute available
Best for: Static sites
Step 1: Get LLM API Key
Sign up with OpenAI, Anthropic, or your chosen provider. Get your API key from their dashboard (looks like: sk-xxxx or similar).
Step 2: Set Environment Variable
On your platform (Vercel/Netlify/Cloudflare), add your API key as a secret environment variable (e.g., OPENAI_API_KEY).
Step 3: Deploy Dendron
Dendron's backend proxy reads ${OPENAI_API_KEY} and forwards requests to OpenAI. Your site calls Dendron proxy, not OpenAI directly.
13KB of behavioral intelligence.
All processing off main thread. <1ms CPU per frame.
Greeting, Feature, Pricing, Engagement, Comparison.
Closed shadow root. CSS never leaks in or out.
Bring Your Own Key. Dendron never touches API keys.
React, Vue, Next.js, vanilla HTML. One API.
Auto-registers actions for agentic browsers.
One script tag. Free hosting. Under 2 minutes.