About honeyp.ai
honeyp.ai is signup protection for products with a free tier. We decide how much free value each new signup should get — full credits for real users, less for the farms, bots, and AI agents gaming it — and link each new account to earlier ones to catch farms. It runs in shadow mode first and fails open, so real users barely notice and an outage never blocks a signup.
Our mission
Let companies stay generous. Free tiers are how good software reaches its next user — but every free credit also tempts farms, bots, and now AI agents signing up on their own. We're building the layer that tells them apart, so a company can give real people a great free experience without funding abuse. We'd always rather reduce and verify than block, because protecting the business should never mean turning away a real person. Catch the farm, not the fan.
Why we started
Every free credit an AI company gives away costs real compute. When one person creates fifty accounts, or a script creates five thousand, that cost lands on a startup's GPU bill. The usual answers are blunt: block more people, demand a credit card, or cut the free tier. Each one hurts the real users the free tier was for.
We think there's a better answer: give each new account the right amount of free value, and explain why.
What we believe
Catch the farm, not the fan
Reduced credits and a quick verification beat a hard block. A real person who looks unusual should still get in.
Show the reasons
Every decision comes with plain reasons, and you can see what we would have done before anything changes.
Collect less
Hashed emails, shortened IP addresses, no keystrokes. Your users' data is never sold or used for anyone else.
Never be the outage
If we're unreachable, your signups carry on. Protection should never cost you a real customer.
The honeyp engine
honeyp.ai began with one question — is this new signup a real person, a farm, or a bot? Underneath the answer is a general decision engine: it turns signals into an honest, explainable risk read and a recommended action — allow, limit, or review — while you stay in control, see what it would do before enforcing, and no real person is hard-blocked for looking unusual. That engine isn't specific to signup.
So the same engine, and the same "catch the farm, not the fan" principle, reach the other grey-area trust calls a company makes — starting with the shift we're building for: AI agents acting on people's behalf.
One engine, many decisions
Signup — live today
Decide how much free value each new account gets, and link the farms. Our wedge, and where honeyp is proven.
Person, bot, or agent
Tell them apart across a product, not just at signup — and give an honest agent a real door instead of making it pose as a human. Agent Check is the first step, live now.
Access & AI resources
The same engine reaches who gets API or platform access, and the abuse of mined keys, free quotas, and farmed compute.
Open about accuracy
We publish how we measure ourselves — our catch rate, and how rarely we get a real person wrong — instead of unprovable claims.
We're winning signup first; the engine is how we grow, and we'll share what we learn as we go.
Who we are
honeyp.ai is built by a former Apple, Google, Meta and Amazon trust & safety engineer, and is operated by Honeyp AI, Inc. We're working with a small group of design partners to shape the product.
Talk to us
Try the live demo, run the free abuse audit on your own signups, or apply for the design-partner program. You can also write to support@honeyp.ai.