In most education strategy conversations, “whole child” appears in the mission statement and the conference title. Then? It gives way to the real discussion about outcomes, platforms, and now artificial intelligence.

Damian Creamer, founder and CEO of Primavera Online School and StrongMind, argues the phrase has been filed in the wrong drawer.

“People treat the whole child as the soft part of education, the part you attend to once the academics are handled,” Creamer says. 

“It is the opposite. Autonomy, competence, and connection are the conditions under which a brain will engage with anything difficult. A system that ignores them is not being rigorous. It is manufacturing disengagement and then going looking for its outcomes.”

The Research Is Not Ambiguous

Creamer’s position rests on self-determination theory, the motivation research developed by psychologists Edward Deci and Richard Ryan. 

Its central finding is that people sustain effort on difficult things when three needs are met, and withdraw effort when they are not.

• Autonomy: some genuine say over how and when the work happens. 
• Competence: visible evidence of getting better at something. 
• Relatedness: which Creamer calls connection: at least one person in the system who knows who you are.

“What makes it useful is that it is predictive rather than descriptive,” he says. “Remove autonomy and you get compliance, which resembles learning right up until you ask a student to apply something in a setting nobody prepared them for. Remove competence and you get avoidance, because nobody volunteers to feel incapable five days a week. Remove connection and the student withdraws, sometimes visibly and more often not. Those are not personality outcomes. They are the documented response to a missing condition.”

Disengagement stops being a student problem and becomes a readout on the environment.

“When a school watches a student stop trying, the instinct is to intervene on the student,” Creamer says. “Attendance plans, credit recovery, a conversation about effort. Sometimes that is right. Very often the student is responding sensibly to a setting that removed one of the three inputs, and no amount of intervention on the child restores a missing input.”

The Conditions Were Priced Out, Not Rejected

If the research is this settled, the obvious question is why so few systems are built around it. Creamer’s answer is that nobody rejected the conditions. They became unaffordable.

“Each one costs the only currency a school really has, which is adult attention,” he says. “One teacher and thirty students cannot support thirty flexible pathways, hold thirty separate trajectories in mind, and sustain thirty real relationships. So the system does the affordable thing. It standardizes, teaches to the middle, holds the pace fixed, and describes the uniformity as fairness.”

He is unsparing about that last move. “Standardization was never equity. It was what we could afford. We built a story around the constraint because we had nothing better on offer, and then several generations inherited the story without anyone re-examining the constraint underneath it.”

The interesting problem is whether the architecture can be rebuilt so the conditions stop being a luxury.

The Test He Applies to AI in Education

Education is in the middle of deciding what artificial intelligence is for. Creamer’s concern is that the industry is directing enormous capability toward what is easiest to scale rather than what meaningful learning actually requires.

“Take any product being pitched to schools this year and ask which of the three conditions it serves,” he says. “A tool that generates more content faster does not create autonomy. A tool that produces an answer on demand does not build competence, and can quietly prevent it. A conversational layer standing between a student and the adult who knows him is working directly against connection while looking helpful in a demo. Volume and speed are easy to show. Neither one was the thing that was missing.”

He is careful that this is not an argument against the technology.

“The honest use of AI in education is to make the expensive conditions affordable,” Creamer says. “That is the opportunity, and it is a big one. Not more material. Not a substitute for the person. Carry the load that made autonomy, competence and connection unaffordable in the first place, so a school can offer all three to every student instead of to whoever got lucky in the scheduling.”

What Designing Around the Learner Requires

At StrongMind, that principle determines where the engineering goes. Creamer draws the distinction deliberately.

“The work does not go into features,” he says. “It goes into a layer underneath the platform whose only job is to make personalization possible at scale. A feature sits on top of a system. This has to be the system, or it does not hold.”

Autonomy means real variation in pace and path. Competence means continuous visibility into mastery, so a student can watch himself improve. 

Connection means the system routes the right signal to the right adult early enough for the adult to act, which turns out to be an engineering problem in service of a human one.

“Technology has a narrow job here and it should stay inside it,” Creamer says. “Carry the load. Do not attempt the relationship. Every serious failure I have watched in this industry came from a product that had those two backwards.”

The Rigor Objection

The standard objection is that all of this trades academic standards for student comfort. Creamer treats the trade as false and answers it with results rather than argument.

Primavera was designated “Highly Performing” by the Arizona State Board of Education this year, and its Class of 2026 passed 1,000 graduates, a large share of them students the conventional system had already lost.

“Not one of them graduated because we were pleasant to them,” Creamer says. “They graduated because the conditions were in place for them to do genuinely hard work over a long stretch of time. Rigor without those conditions is not rigor. It is a filter, and what it mostly sorts for is how much support a student happened to have at home.”

A Specification, Not an Aspiration

What separates Creamer’s version of this argument from the mission-statement version is where he puts it in the sequence. 

The whole child is the specification the system must satisfy before the academic goals become reachable.

“Reverse the order of the conversation,” he says. “We design the delivery mechanism, then ask how to keep students engaged with what we built. Start instead from what a person requires in order to learn something hard, and build the thing that can supply it to a million of them at once. The second half is genuinely difficult. It is also the actual work, and it is the part that AI has only just made possible.”

Twenty-five years in, Creamer does not present that as finished. He presents it as the reason the next few years matter more than the last twenty.