Product
Humanizer
An editorial pipeline that makes AI drafts read like a person wrote them
LLM pipeline · Scored QA loop
The problem
A model can draft a page in seconds, and a reader can smell it just as fast. In health, legal and finance the default machine voice is worse than awkward: it makes claims a compliance reviewer would strike.
How it works
Every draft goes through a loop. Write, score it against a catalog of machine-writing tells, rewrite, score again. Content gets classified by industry first, so a treatment center page is held to different rules than a product description. Nothing ships until the score clears a floor we set by scoring draft after draft until the numbers stopped moving.
It runs quietly behind the copy work on our client sites.
More of our work
Product
Simple
Capture a task in one line. Simple scores it on time, urgency and importance, then sorts the day into what to do now, what can wait until this week, and what can wait.
Product
Build
A native macOS and Linux app that runs the real Claude Code and Codex CLIs in true PTYs, across local and remote machines, in one window with a project-scoped browser beside them.