A surgeon with no time to write now approves articles built from the evidence instead of generic health copy.
The draft is scored by an agent that had no hand in writing it.
The situation
How this used to work
A practising orthopedic surgeon has the two things medical content needs and never has at once: the clinical authority to say something true, and no time to write it.
The alternative is the one every clinic takes, which is generic health copy assembled by someone who is not a doctor.
What we built
The system
An article pipeline running as a team of five agents on the Liberators OS, each owning one stage: research, content strategy, the article, an evaluation pass over the result, then publishing.
The evaluation stage is the one that matters. An agent that writes and then approves its own writing is not doing quality control, it is justifying itself. The pass that scores the draft is a different agent from the one that wrote it.
The output is a Google Doc. A doctor reviews and signs off on a draft. Nothing publishes itself.
How it works
The pipeline, step by step
Find what is actually known
The pipeline opens with research rather than with a topic, so the article is built from what the evidence supports.
Decide what the piece is for
A content strategy pass sets the angle and the shape before a word of the article is drafted.
The draft
The article gets written against the strategy and the research from the first two stages.
A different agent reads it
The draft is scored by an agent that had no hand in writing it, before any human sees it.
Into a doc, for a human
The result lands in a Google Doc. The surgeon approves it. That gate does not move.
What changed
The result
Delivered on 16 June 2026 and in testing with the client.
It runs on the same Liberators OS agent teams every other client's system runs on. There is no bespoke stack behind it.
The stack
Where it runs
- Liberators OS
- agent teams
- agent-scoped knowledge base
Output
- Google Docs