Programming systems for Human / AI collaboration.
Software has always made humans more productive. Now there's a new computing primitive — LLMs — and the potential for a real jump is obvious. What's not obvious is the interaction model.
Chat was the first step, and it's being bolted onto our last-generation tools. But what does it look like when you put a reasoning primitive at the core and build new tools from there?
A programming framework that gives AI the structure it needs to be genuinely useful on hard domains. So humans and AIs can align on the "what and why," not just discussing the output.
A developmental editor for novelists, built on creAItive. It extracts the structure — themes, character arcs, story beats, pacing — and lets authors work at that level, not sentence by sentence.
We’re distilling the experience of a professional ML team into a simple programming interface for you and your coding agent. Reliability, cost tracking, user feedback, and quality evaluation work together, so you can build and improve an intelligent application without reinventing all the machinery behind it. The companion book explains the thinking behind the code.
One judgment interface, six ways to execute it. Comparing Jev, OpenAI, Claude, and local DeepSeek on quality, cost, and latency.
Coding agents have a golden path. What happens when building an intelligent system takes them off it—and can an API carry some of the experience they’re missing?