Give each part a clear role.
A human creates the workflow and sets the direction. AI participants take on explicit responsibilities within it.
Human direction. AI collaboration.
Different minds.
Shared direction.
An evolving way to work with distinct, interchangeable AI participants — connected by a common structure and guided by a human.
Explore the approach01 / THE ARRANGEMENT
DIRECTION + JUDGEMENTHumanDefines the work.
Reviews what comes back.
Reviews what comes back.
AI PARTICIPANT / 01Dot1
AI PARTICIPANT / 02Hermes
Parts can change.
The work stays connected.
INITIAL PARTICIPANTSEVOLVING BY DESIGN
KJL is the name. The work gives it meaning.Ideas into practice / 01
01 / The approach
The way they
work together
is the point.
Adding more AI is only a starting point. KJL explores how separate participants can contribute to one coherent piece of work.
Make the handoff legible.
Shared formats, written decisions and durable records help the next participant pick up the work with its context intact.
Check the work. Adapt the structure.
Review evidence and results. Draw on proven ways of organising work, then adjust responsibilities and replace parts as needs change.
02 / In practice
A thread through
the whole process.
The aim is continuity: a short request becomes reviewable work, and what is learned remains useful beyond a single conversation.
Start a conversation.
Short requests and links keep coordination close to the work.
SlackKeep the context.
Knowledge and working records connect across notes and versioned material.
Obsidian + GitHubMake it inspectable.
Briefs, results and artifacts give the human something concrete to review.
Drive03 / A foundation to build on
A practical approach.
Still taking shape.
KJL begins with OpenAI models, ChatGPT and Codex as its chosen foundation. Dot1 and Hermes are the initial AI participants. The workflow evolves through use, review and deliberate changes.
KJL is an independent project. It is not affiliated with or endorsed by OpenAI.