Recent articles 1 – 9 of 49
Should you be nice to your AI?
Not because the model might feel it. Because of what it does to the answer, and to you.
Does your chatbot have to let customers reach a human?
The condition your deflection target can quietly break.
The boundary is the thing you can draw
Why judging each request is the wrong shape, and what to do instead.
A watermark is not a disclosure
Two obligations, one announcement, and the gap between them.
Model access is a supply chain
Your model provider now has a regulator, which makes it a supplier like any other.
The cease and desist I was secretly proud of
Years before conversational AI was a thing, we were changing famous websites to demo an idea. Their security caught us before a human did. The letter that arrived was polite, entirely deserved, and the first real sign the idea mattered.
What your agent remembers, and for how long
Four questions about agent memory, and what a good answer sounds like.
Confidence percentages are theater
A number next to an answer feels like evidence, and almost never is.
How we got our name (after a Klingon detour)
We burned through two names before this one. A trademark we did not check, a domain we could not afford, and a country code with a reputation. The fix was not inspiration, it was a scoring method, and the discipline to walk away from the name we loved most.