The project asks how to do knowledge work well with AI agents. Its first commitment is to useful work and transparent inquiry.
Six questions to begin with
- What are we trying to achieve or learn?
- What do we currently know?
- What can the agent do independently?
- How will the work become inspectable?
- How will we judge whether it is useful?
- What should future work learn from this?
Keep the practice proportional
For a small assignment, six sentences may be enough. A recurring process may need shared instructions, preserved evidence, and repeatable checks.
Still to investigate
Which practices transfer across different kinds of work? When does structure become overhead? How do we preserve discovery? What does systems thinking add?
These are starting hypotheses to examine through research and practice. Future versions should link recommendations to the cases that support them.