What is the Brainfulness Cockpit?
The Brainfulness Cockpit is a four-layer framework for working with AI: the flight plan (your values, purpose and meaning), the pilot (you), the copilot (the AI) and the decision (who is in command and what an error costs). It has one rule: the copilot never lands the plane.
It is a practitioner framework by Hans de Graaf, built on cited research. It is not a scientific theory. Where a part rests on a specific study, the source is given.
What are the four layers?
Flight plan: why are you flying at all?
The flight plan is your values, your purpose and what makes the work meaningful. It is the layer most often skipped when tasks are handed to AI. A meta-analysis of 44 studies with 23,144 workers found that meaningful work goes together strongly with engagement and commitment, and with lower intention to leave (Allan et al., 2019). Hand the meaningful part of a job to AI and you may keep the output and lose the person.
Example: an agency owner lets AI write all her client proposals. Output doubles and she stops enjoying the work: writing was where she felt she made a difference. Once she names craftsmanship as a core value, she knows which part of the job to keep in her own hands.
Pilot: what is your own brain doing?
The pilot is you. Your brain reacts before you have thought things through, matches patterns that may not fit, and builds convincing stories from incomplete information. The I5 Brain Model describes five functional domains, five instruments on the pilot's panel. Each gives a reading; none should fly the plane alone. You cannot feel your own miscalibration (Fernandes et al., 2026), so you need a routine that catches it.
Copilot: how far can you rely on this tool, for this task?
"Do I trust AI?" is the wrong question. Three better ones:
Able. Is this tool good at this task? AI combinations tend to help with creating content and to hurt on decisions (Vaccaro et al., 2024).
Motive. Whose interests is it optimized for? Expect agreement, not truth: AI affirms users 49% more than humans do, and users prefer it that way (Cheng et al., 2026).
Stable. Is it consistent, and does it say when it does not know?
These three questions come from the best-known model of trust between people (Mayer, Davis & Schoorman, 1995: ability, benevolence, integrity), translated so that they stop pretending the tool is a person.
Decision: who is in command?
The decision layer is about command: who owns the result, what an error costs, and who may say "I'm not comfortable with this." Aviation introduced crew resource management after accidents in which a crew member saw the problem and did not speak up. With AI, the copilot never hesitates, and the voice that stays silent is often your own. In the cockpit everyone may speak. Only the pilot signs.
What do you actually do? HOLD.
HOLD is a four-step cross-check for working with AI: Hands off, Own view, Look at the copilot, Decide. It takes 5 seconds, 30 seconds or 5 minutes, depending on what an error would cost.
What about the organization? The Tower and the Airline.
The Tower is the human governance of AI in an organization: leadership, values and ethics. The Tower uses AI tools. It is never run by them. A pilot-in-command may refuse a tower instruction on safety grounds, and that is the point.
The Airline is the whole organization seen through the same metaphor: pilots, tower, ground crew and passengers. Every seat in the airline has AI in it. Only one seat has AI as a colleague.
What is borrowed and what is new?
Most of the building blocks are not mine, and I would rather tell you than have you find out. The values-purpose-meaning chain draws on research into meaningful work (Steger, Dik & Duffy, 2012; Allan et al., 2019) and on Amabile and Kramer's progress principle. The fast and deliberate sides of the I5 model echo dual-process thinking. "If X happens, I will..." is Gollwitzer's implementation intention. The copilot questions translate Mayer, Davis and Schoorman. The decision layer is crew resource management.
What is mine: the I5 mnemonic, the cockpit as the frame that holds it together, HOLD as a ritual for the moment you work with AI, the three gears, and the examples.
Brainfulness
Brainfulness is a practitioner framework by Hans de Graaf for understanding how your brain shapes your thinking, decisions and behavior, and for deliberately working with it instead of letting it fly on autopilot. It is built on cited research. It is not a scientific theory and it is not a form of meditation.
The I5 Brain Model
The I5 Brain Model describes five functional domains of the brain: Indispensable (keeping you alive), Instinct (being human), Intuition (learning from experience), Intelligence (thinking deliberately) and I (being you). It is functional, not anatomical: it is not a map of brain structures and not a ladder from an "old" to a "new" brain.
The Brainfulness Cockpit
The Brainfulness Cockpit is a four-layer framework for working with AI: the flight plan (your values, purpose and meaning), the pilot (you), the copilot (the AI) and the decision (who is in command and what an error costs). It has one rule: the copilot never lands the plane.
HOLD
HOLD is a four-step cross-check for working with AI: Hands off, Own view, Look at the copilot, Decide. You use it before you hand work to AI, when something goes wrong, and afterwards. It takes 5 seconds, 30 seconds or 5 minutes, depending on what an error would cost.
The Tower
The Tower is the human governance of AI in an organization: leadership, values and ethics. The Tower uses AI tools. It is never run by them.
The Airline
The Airline is the whole organization seen through the same metaphor: pilots, tower, ground crew and passengers. Every seat in the airline has AI in it. Only one seat has AI as a colleague.