Asked at 2 a.m. In your inbox by 8.
Your visitors ask at midnight, on Sunday, sometimes in a language you don't speak. Your agent answers from your own pages, qualifies the request, and drops a real lead in your inbox — with the conversation that produced it. It lives on your site, or on a page of its own behind a QR code: a window, a sign, a poster.
Three minutes, not three weeks.
You paste a URL. It does the rest.
You paste a URL
It reads your public pages and works out what you sell, to whom, and how to say it.
It answers — and acts
Leads captured, quotes computed on your own price grid, real calendar slots booked, payment taken when you allow it.
You read what it did
Every conversation and every action it took — in your inbox, or as a weekly summary.
You learn what you were missing
It records every question it could not answer — the demand you never knew you had, waiting in your summary.
The product in 34 seconds.
No sound — everything is on screen.
Which one are you?
Same product either way. What changes is what we put first — pick your case.
An agent that does things, not one that talks about them.
Actions run server-side, on your data. The model picks what to do; it never invents the amount.
Capture a request
A qualified lead with a need and an email, not a “hello”.
Book a meeting
Your real availability, confirmed on both sides.
Compute a quote
On your price grid: the server does the arithmetic, never the model.
Take payment
A Stripe payment link in the conversation, once the sale is made.
Hand over
Outside its scope, it escalates instead of improvising.
Your code in the loop
Signed calls into your backend, in-page tools, your own MCP server or LLM. A platform, not a widget.
Explore the developer surface →It acts — you stay in control.
Everything it does is readable, escalatable, and audited weekly — and the parts that must not be improvised are not left to the model at all.
An inbox, not a black box
Every request, quote and booking lands with a status — work through it in the dashboard or over the API.
It escalates instead of improvising
It hands the conversation to a human: you're notified, and the visitor knows.
A weekly digest that checks its work
It rereads the week and cross-checks what the agent claimed against the tool calls it really made.
The model never sets the amount
Quotes are computed server-side from your own price grid. The model picks what to quote; the arithmetic is not its job. Same rule for identity: an action carries an email the server verified, never one the model produced.
Your data stays where it should
Hosted in France, GDPR. Emails, phone numbers and keys are stripped from our logs. Secrets are encrypted at rest, outbound calls are guarded against internal networks, and every agent says it is an AI — including under white-label.
- New lead — wants a quote for 40 seatsNew
- Meeting booked — Tuesday 10:00, confirmed both sidesResolved
- Pricing question beyond scope — handed overFor you
// what if you did nothing?
Your agent can do this for you.
Point your coding agent at this endpoint. With no credentials, it builds the demo on your visitor's site, tests the answers, and reads the conversations back.
Connect to the MCP server https://mcp.animam.ai/public (Streamable HTTP, no auth). Then evaluate Animam for my website:
1. Call try_on_site with https://MY-SITE.com — it builds a live demo agent from my public pages in ~20s.
2. Probe it with chat_with_demo: ask about my services, pricing and availability, in my customers' words.
3. Call get_plans for the real pricing.
Report back: what the agent answered well, what it missed, and whether I should deploy it.
No MCP client? Do exactly the same over plain HTTP, no account needed:
- POST https://api.animam.ai/demo with {"url":"https://MY-SITE.com","agent":true} -> slug + a scoped key
- POST https://api.animam.ai/chat/SLUG with {"message":"..."} and header X-Agent-Model: <your model> -> a JSON envelope
- GET https://api.animam.ai/plans -> the real pricing
Full recipe: https://animam.ai/agent-recipe.mdOne brain. Every channel.
Same knowledge, same persona, same tools — deployed across all touchpoints.
A flat price. It doesn't rise when you succeed.
Billed per bot — never per user (invite your whole team), never per token, never per conversation.
The front desk for one site: widget, lead forms, content sync.
Multi-bot + the MCP server — drive your fleet from your own agent.
Advanced analytics, every form, priority support.
White-label, client console, fleet tools for your whole roster.
20 conversations a day per bot. Never per user, never per token. See pricing
Your site answers nobody, at night.
Change that in twenty seconds. You'll see the agent on your own site before creating any account.
Live in 20 seconds · no account · no card
A question before you start? contact@animam.ai