What broke,
and what the measurement said.
Write-ups from building this alone. No case studies, no accuracy figures, no customer logos, because there are none of the three and inventing them is the one thing that would make the rest worthless.
What is here instead: things the product got wrong, the number that proved it, and the change that fixed it.
An API that silently ignores a field is lying to you.
The one person who ever evaluated this API sent a field name I do not accept. It was ignored without a word, they were told they looked automated three times, and they left. The wrong name came from my own example.
Anatomy of a 55,000-bot attack on an anti-bot startup.
We sell bot detection. I opened the database and found 555 pages of fake signups. What broke, why it is embarrassing, and the accidental circuit breaker that saved the email domain.
The page that lists what it does not catch.
Where useHUMA is the wrong tool, where the signal does not exist, and where it stops on purpose. Including the cases where the honest recommendation is a competitor.
Read the limitsTry the demo