What separates a mind
from a model?

We treat every bot as a black box — and study it from three angles: the framework fingerprints of the stack it runs on, the biological perception it cannot fake, and the behavioral reasoning patterns it leaks. From that research we design CAPTCHAs, device fingerprints, and detection systems that tell humans and bots apart.

Explore Products Read the Research

99.4%

Bot detection rate

12ms

Median verification

4.2B

Signals analyzed / mo

0.7%

Human friction rate

⏤ What we do

Three layers of human verification

01 / CAPTCHA

Cognitive Challenges

Tasks designed around human-only perceptual reasoning — hard for state-of-the-art LLMs and vision models, trivial for people.

02 / SHIELD

Anti-Bot Infrastructure

Continuous behavioral telemetry: micro-mouse dynamics, scroll cadence, attention shifts. Bots fail silently, users feel nothing.

03 / RESEARCH

Cognition R&D

We publish primary research on the perceptual gap between brains and transformers — the science our products are built on.

⏤ Studying AI for Good

We learn from AI
to protect humans

Every frontier model that learns to fool a CAPTCHA teaches us something about its blind spots. CogBio runs a continuous adversarial program against GPT-class agents, multimodal models, and headless browser swarms — feeding insights back into open research and stronger defenses.

Adversarial benchmarks
Public datasets that quantify the human–AI perceptual gap.
Responsible disclosure
We coordinate with model providers when we find new weaknesses.
Open research
Quarterly reports on how AI capabilities reshape the bot-defense landscape.

Visit our research lab

⏤ Recent updates