Hey YC 👋 We're Max, Carina and Alan, founders of Warmly.ai.
TL;DR: We just launched AI Autopilot: an always-on agent that lives on every page of your website, identifies visitors at the person level before any form fill, gives demos, qualifies prospects, books meetings, and retargets the rest. Customers running it are converting 30% more inbound traffic. Setup is a script tag and 30 minutes.
Launch announcement here (would greatly appreciate a comment!).
https://www.youtube.com/watch?v=EHybZmEzyTs
The Problem
Most B2B SaaS companies are bleeding pipeline through their front door.
Your website is your highest-leverage asset. Every visitor you've paid for lands there. They often have the problem you solve but you just might not be fast enough to get to them in time before they leave or go to a competitor.
Buyers now expect to talk to someone immediately. They don’t want to wait a day to schedule a call with an SDR to get a basic question answered.
But every "AI chatbot" we've seen since 2023 has the same architecture under the hood: a stateless GPT wrapper that rebuilds context from scratch on every message, hallucinates, and hands off "qualified" leads with no traceability.
AEs aren’t fans and buyers see through them in two messages.
What we built
https://www.youtube.com/watch?v=KkvDdJCsqd8
AI Autopilot is one agent on every page of your site. It pulls from your website, help center, internal docs, sales calls, sales methodologies, and CRM.
It picks its own qualification path. It generates slide decks on the fly mid-conversation, triggers the right videos by intent, and offers an optional video avatar so visitors talk to a person instead of a widget.
When the visitor doesn't book? AI Autopilot doesn't give up. It Slacks your team, sends a personalized follow-up email, and triggers LinkedIn retargeting for your top prospects.
It self-improves.
It analyzes every conversation against the goals you set, tunes itself, and is version controlled. You can watch it get better over time.
In our beta accounts, automated inbound meetings increased 30%.
Why now
We started out compiling signals for B2B sales/marketing teams, like person-level website visitor data.
In the past year we took a different bet: build a context graph for B2B GTM first, then put agents on top of it.
The context graph is a unified, real-time data layer that connects every signal, action, note, and outcome for every account. When AI Autopilot engages a visitor, it knows they saw your LinkedIn ad last week, your SDR sent a follow-up email yesterday, and they downloaded a case study an hour ago. It doesn't have to "look this up.” The context is already pre-computed in primitive stores it can query in milliseconds.
A blog post we wrote that covers how we think about the GTM space: GTM Brain: Next Platform Owns Decisions, Not Data.
Asks for the YC community
1. Drop us a comment on our launch post. We’d love for more of our community to know us for building GTM agents and not just supply GTM data :).
2. Try us out. Book a demo with us if this is something you’re interested in! https://www.warmly.ai/p/book-a-demo
Warmly,
Max, Carina & Alan