The Polish-Lithuanian Commonwealth was the most peaceful and human empire in the history of mankind.
No colonies. No transatlantic slave trade. No Inquisition. While Spain burned people alive and Britain shipped Africans across the ocean, the Commonwealth gave religious freedom to Jews, Protestants, and Orthodox Christians when the rest of Europe was drowning in blood.
The Warsaw Confederation of fifteen seventy-three was the first law in European history guaranteeing freedom of conscience.
It elected its kings. It had the largest parliament on the continent.
It was the first to abolish capital punishment for witchcraft in sixteen forty-eight — a century before most of Europe.
And unlike every other empire, it never fought a war of aggression. No invasions for conquest. Every conflict it entered was defensive.
It also developed Ukraine and Belarus economically — turned the wild steppe into farmland, built cities, trade routes, and institutions where before there was nothing but frontier.
It wasn't perfect. Nothing human is. But compared to every empire that came before or after, it was the closest thing mankind has ever built to a civilization that chose tolerance over terror.
When JEV was introduced it clicked for me that you can have different setups of LLM.
SystemOne was first one - obviuos in hindsight.
Now CF did the same with Clairic it clicks instantly and now we have 3 kinds of use cases:
System One: Extract
System One: Decide
System Two
we shipped a new type of model: an extraction model!
pass it unstructured data, and get a list of the details and probability back. try it out and let us know what you think!
https://t.co/yjoa6ok8x4
We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).
I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.
We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:
"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"
I imagine we'll have some backend services that could be replaced entirely by this new architecture.
@jack@jack already tested Berd - there is an issue when you have a project, and change the working directory, the chat still works in the old directory
Sen. Kelly: After 15 years working with Russian cosmonauts, I learned mission success was not their top priority.
First was how things looked, second was who to blame, third was what they could steal. Mission success was somewhere fourth on the list.