Human
You choose what matters, give consent, accept commitments and risks, and decide when to stop.
Purpose before performance
Know what matters. Then learn how to move it.
There is plenty you could do. What deserves your time and care? Start with who you are, what matters to you, and the future you want to help create. Once the direction is yours, the open Playbook offers practical know-how for the next step.
Dreamineering helps people and AI agents work together on what matters. You own the purpose, consent, and decision to stop; an agent works within the bounds you set.
From direction to action
Purpose sets the direction. The Playbook equips.
What you value shapes what you notice and choose. Use Value, Belief, and Control to hold a course, then choose a Playbook recipe for one bounded move in your AI-native venture. Review what changed to improve the next choice.
- 1
Purpose
- 2
Value · Belief · Control
- 3
Playbook know-how
- 4
One bounded move
- 5
Evidence
- 6
Better next move
Human · Agent · Together
Coordinate capability without giving away authority.
Human authority stays with the people responsible for the work. People affected retain their consent; an agent's ability to act does not give it permission.
Agent
An agent can research, compare, prepare, and act within the permissions you give it. Ask it to keep evidence for your review.
Together
Agree who the work should help, who does what, and what would count as a worthwhile change.
What is here
A guide, a developing substrate, and public experiments.
Use the guidance, follow the development, or inspect an experiment. Each has a different role.
- Dreamineering
- The open guide and Playbook for practical decisions and action, with evidence improving the next choice.
- Stackmates
- The coordination substrate—the supporting software for people and agents—being developed and tested. Its capabilities remain subject to evidence and human permission.
- Ideas
- Venture experiments where we try the guide and developing software in our own work. An experiment's existence does not prove it has helped anyone.
Learn from what changes
Did the work make a worthwhile difference?
A completed task alone cannot answer that. Look at what changed for the person the work should help, including time spent waiting. Check the effect alongside costs, errors, and workload, then use the evidence to keep, revise, or stop the next move.
Start with the work in front of you
Choose the smallest recipe that fits.
Name who the work should help and the job to do. Choose one step within your permissions, then bring back evidence that can improve the next move.
Open the Playbook