intentLM creates a universal grammar for customer behavior
The real-timeintent layerfor product behavior.
One installation. Three consumption paths — teams, models, and agents. Classified while users browse. No PII leaves the browser — only integer token IDs.
// user navigates — intentLM classifies in background
0ms added at response timeTry the live sandbox
No signup. Click around a fake product and watch the intent panel classify your behavior.
FREE — 10,000 sessions free every month
Start on the free plan — 10,000 Monthly Active Sessions (MAS), web app only, no credit card required. A session tracks a user's continuous interaction stream up to 30 minutes of inactivity. Connect your repo, map URLs to tokens, and upgrade automatically when your app scales past the limit.
Create account →One signal layer. Three ways to use it.
intentLM turns raw product behavior into structured intent — then routes it to whoever needs to act, plan, or model.
Humans
Push intent signals to dashboards, Slack, and your agentic layer so CS, Sales, and Product teams plan from live behavioral signal.
Predictive systems
Export stable token sequences, intent classes, confidence scores, and cross-session trajectory as clean inputs for churn, expansion, and LTV models.
Agents
Emit a real-time Intent Object into your AI agent's context before the first message — so support, sales, and copilots act on classified intent immediately.
Agents & MCP
One of three consumption paths: connect your AI agent to intentLM via MCP (Model Context Protocol). Classify user intent, configure patterns, and trigger actions from your agent's tool calls.
12 tools
analyze_session, set_url_pattern, suppress_intent, and more
4 prompts
setup-intentlm, diagnose-intent, configure-alerts, review-coverage
Works with
Claude Code, Cursor, any MCP-compatible agent
// Claude Code — .claude/settings.json
// Cursor — Settings → MCP, or .cursor/mcp.json
{
"mcpServers": {
"intentlm": {
"command": "npx",
"args": ["-y", "@intentlm/mcp-server"],
"env": {
"INTENTLM_API_KEY": "ilm_live_...",
"INTENTLM_CONFIG_URL": "https://intentlm-dev-config-krxe5fa7dq-uw.a.run.app",
"INTENTLM_INFERENCE_URL": "https://intentlm-dev-inference-krxe5fa7dq-uw.a.run.app"
}
}
}
}FREE — 10,000 sessions free every month
10,000 Monthly Active Sessions at $0
No credit card required. Web app only. MAS = a continuous interaction stream up to 30 minutes of inactivity. Markov model, 1 webhook, community support.
Create account →What is intentLM?
intentLM is a real-time behavioral intent classification platform for SaaS, B2B software, and e-commerce. A lightweight browser SDK captures navigation and product interactions, converts them into a shared global taxonomy of integer token IDs, and sends only those abstract tokens to the inference API — never raw URLs, form values, or personally identifiable information.
Product teams use intentLM to classify live sessions into intent classes such as upgrade seeking, churn risk, onboarding friction, and support escalation. Customer success and sales teams route signals to Slack and webhooks. Data teams export token sequences for churn and expansion models. AI agents connect through the MCP server to inject classified intent into agent context before or during a conversation.
Get started with the live sandbox, read the integration docs, or create a free account.