Powered by MindLevr™
Introducing MindLevr™ Adaptive Training Engine and Rep Builder 10 — a new generation of training systems that adapt to the athlete instead of forcing every athlete to follow the same path.
By David Brannan ·

Two athletes start the same pull-up program on the same Monday. Six weeks later, one of them hits their goal ahead of schedule. The other stalls somewhere around Week 3, keeps grinding through workouts that are technically still “on plan,” gets frustrated, and quits by Week 5.
Same program. Same instructions. Two completely different outcomes—and nothing on the page ever noticed the difference.
Any coach who has spent real time on a gym floor has watched some version of this happen more times than they would like to admit. You learn to recognize the athlete who is progressing faster than the program expected and eventually starts getting bored or sloppy. You also learn to recognize the athlete whose numbers on paper look fine, but whose performance is quietly telling you that something is no longer working. A good coach adjusts. That is a large part of coaching, honestly—not simply writing the plan, but recognizing the moment when the plan stops fitting the person doing it.
A printed program cannot do any of that. It just keeps going.
That gap is something I have spent decades thinking about as both an athlete and a coach: how do you give someone enough structure to make meaningful progress without losing the flexibility that real people—with real bodies, changing abilities, unpredictable schedules, and occasional bad weeks—actually need?
It is a harder question than it sounds.
Why We Set Out to Build an Engine, Not Another Program
Writing a training program is relatively easy if you assume everyone will respond to it in roughly the same way. Pick a progression, assign the workouts, tell the athlete exactly what to do each day, and move through the schedule. That is not a criticism. Some of the best training systems ever created work exactly this way, and they have produced remarkable results.
The problem is not that these programs are bad. The problem is that the athlete on Week 8 is no longer the same athlete who started on Day 1. Their strength has changed. Their work capacity has changed. Their confidence may have changed. Their recovery may have improved—or worsened. Life outside the gym may have interrupted the schedule. Some athletes adapt faster than expected, while others need more time than the original program allowed for.
The training plan may have been written once, but the athlete continues changing.
Much of the training software available today has been designed primarily to manage the plan rather than understand the person following it. It stores workouts, calculates percentages, tracks completion, produces charts, and reminds the athlete what comes next. Those are all useful functions, but they generally begin with the same quiet assumption: the program is already finished before the athlete ever shows up, and the athlete’s job is to follow it.
We wanted to approach the problem differently. Instead of asking, “What should every athlete do next?” we wanted the software to ask the question a real coach is constantly asking:
What does this athlete need next, based on everything that has happened so far?
That question is where MindLevr™ came from.
Where MindLevr Actually Started
Here is the part of the story that most people do not know: MindLevr did not begin as a training tool at all.
Years earlier, we built the underlying concept as part of eProfessor.com, our learning management system. The problem had nothing to do with pull-ups, squats, or strength training. It was about learning. Every student who enters a course is slightly different from the student who entered it yesterday. One student is ready to move faster than the syllabus allows. Another needs the same concept explained a different way. One is coasting through material that is too easy. Another is quietly falling behind but may never raise a hand and say so.
A fixed curriculum does not see any of that. A good teacher does.
The more we thought about that problem, the more obvious the parallel became. Education and athletic training look very different on the surface, but the underlying challenge is remarkably similar. You meet the person where they actually are. You give them the next appropriate challenge. You observe what happens. Then you adjust what comes next based on the result rather than blindly following a plan that was written weeks or months earlier.
That is what a good teacher does. It is also what a good coach does.
Once we saw the connection clearly, the question stopped being, “Should we build an adaptive engine for training?” The better question was, “Why would we build an entirely separate approach when we had already spent years working on the same fundamental problem in another field?”
Bringing MindLevr into athletic training was not simply a rebrand or a marketing decision. It was the recognition that learning and training are both adaptive processes. In both cases, the person changes because of what happened previously, and the next step should reflect that change.
How MindLevr Actually Works
MindLevr™ Adaptive Training Engine is the technology layer that allows a training program to respond to the athlete as the program unfolds.
It starts with information about where the athlete is now: current performance, selected exercises, training frequency, goals, and other program-specific inputs. As training begins, it adds new information from completed workouts. It can consider what was prescribed, what was actually completed, whether the workout felt appropriate, whether performance is improving, whether repeated difficulty is developing, whether training was skipped or rescheduled, and how all of those signals fit into the larger progression.
Underneath that experience is a deterministic decision engine. MindLevr is not simply pulling a random workout from a library, and it is not using the phrase “AI-powered” as a substitute for having an actual training methodology. Each program still has its own defined rules, progression model, limits, and coaching philosophy. MindLevr uses those rules together with the athlete’s evolving state to determine the appropriate next prescription.
That distinction matters.
Adaptation does not mean endless variation. A good coach does not change a program every day simply because they can. Most of the time, successful training should continue because it is working. The system needs to know when to leave the plan alone just as much as it needs to know when to change something.
MindLevr therefore looks for patterns rather than overreacting to a single workout. If an athlete has one unusually difficult day, that does not automatically mean their ability suddenly disappeared. If several similar sessions begin showing the same problem, that becomes more meaningful evidence. Likewise, a workout feeling easy does not automatically mean the athlete’s estimated capacity should suddenly jump. The system is designed to combine objective performance with patterns across multiple sessions before making larger adjustments.
That was one of the most difficult parts of building the engine. It is relatively easy to make software that changes workouts. It is much harder to make software that knows when a change is actually justified.
The goal was never to make training more complicated. The goal was to make it more responsive while preserving the underlying logic that makes the program effective in the first place.
Rep Builder 10: The First Program Powered by MindLevr™
Rep Builder 10 is the first training program built entirely around the MindLevr Adaptive Training Engine.
The idea behind it is straightforward. Bodyweight strength movements are some of the most accessible measures of physical ability available. Pull-ups, push-ups, dips, bodyweight squats, and core movements require relatively little equipment, but they reveal a great deal about strength, endurance, control, and work capacity.
The difficulty is that athletes arrive with an enormous range of abilities. Someone working toward their first strict pull-up needs a completely different progression than someone who can already perform 25. An athlete trying to move from 20 push-ups to 50 should not necessarily follow the same path as someone trying to move from 50 to 75. Even two athletes who begin with the same maximum may respond differently once training begins.
A traditional program can provide both athletes with a reasonable general path. An adaptive program can respond to what happens to each athlete while they are following that path.
That is what Rep Builder 10 is designed to do.
The athlete chooses the movements they want to improve, establishes a starting point, defines a goal, and chooses how frequently they can realistically train. The program begins with an actual performance baseline and builds training around that verified ability rather than assuming the original estimate was perfect.
From there, the engine manages more than a simple increase in repetitions. It balances training volume, the size of individual sets, training frequency, exercise stress, and where the athlete is within the 10-week progression. Early training emphasizes building quality work. Later training gradually shifts toward larger uninterrupted efforts. Volume eventually comes down as the athlete moves toward the final test.
The program also has to work across very different levels of ability. An athlete who can perform only one or two pull-ups may need strict repetitions mixed with assisted work. An athlete who can perform 50 push-ups should not simply receive endlessly larger sets because a percentage formula says so. High-capacity athletes need different safeguards than beginners, just as duration-based movements such as planks cannot be treated mathematically as if seconds and repetitions were interchangeable.
Those differences are built into the program rather than being left for the athlete to figure out.
As real training occurs, MindLevr evaluates the results and updates the athlete’s current training state. If the athlete is consistently handling the work well, future prescriptions can gradually reflect that improvement. If similar workouts repeatedly become difficult, workload can be adjusted before assuming the athlete has actually lost ability. Missed sessions are treated as scheduling events rather than automatically being interpreted as physiological failure. If an athlete changes the days they can train or even changes training frequency during the program, future training can be recalculated without rewriting what already happened.
That last point is important because history matters. A completed workout should always remain exactly what it was when the athlete performed it. MindLevr adapts the future; it does not rewrite the past.
By Week 10, the athlete performs the same type of maximum test used to establish the original verified baseline. The result is then compared with where the athlete actually began, the 10-week target, and—when appropriate—the athlete’s longer-term goal.
The program is not simply marching down a predetermined chart. It is applying a defined training methodology to an athlete whose abilities and circumstances are changing throughout the program.
Why This Was Never Just About One Program
Rep Builder 10 matters, but it was never the larger goal. It is the first demonstration of what becomes possible once the adaptive engine exists.
Building one good training program is valuable. Building the technology that allows many different programs to respond intelligently to the athlete is something else entirely.
That is why we spent so much time building the engine beneath Rep Builder rather than simply writing ten weeks of workouts and calling the program finished. The work involved creating ways to represent an athlete’s changing capacity, generate sensible prescriptions, evaluate actual training results, respond conservatively to success or difficulty, handle missed and rescheduled sessions, preserve historical workouts, and continue generating the program from the athlete’s current state.
Once that infrastructure exists, we do not need to solve those problems from scratch every time we create another adaptive program.
Future AthleticPeak Pro programs can be built on the MindLevr foundation while still having completely different methodologies. A strength program might adapt load and volume differently from Rep Builder. A speed program might evaluate sprint performance, recovery, and exposure frequency. A jumping program could respond to approach speed, jump performance, workload, or training readiness. A general athletic-development system may combine several types of performance data.
The specific coaching rules will change from one program to another. The larger architecture does not have to.
The athlete trains. The system records what actually happened. The program interprets that information according to its methodology. The next prescription reflects the athlete who exists now rather than the athlete we assumed would exist when the program was first written.
That is the larger problem MindLevr is designed to solve.
Where This Leads
For most of the history of training programs, there has been an unavoidable tradeoff. A written program can reach thousands of athletes, but it cannot know any of them. A good coach can know an athlete extremely well, but one coach can only work closely with a limited number of people.
Adaptive training software sits somewhere between those two worlds.
It will never replace the experience, observation, intuition, or relationship of a great coach, and that is not what we are trying to build. A coach sees things software may never see. But there is a large amount of coaching logic that can be defined, tested, applied consistently, and improved over time. That creates the possibility of giving far more athletes access to training that responds to them instead of simply handing everyone the same sheet of paper.
That is the part of this project that excites me most.
Rep Builder 10 is a new program, but the more important development is what now exists underneath it. We have an adaptive training engine that can become the foundation for the programs we build next.
Much of the work now is the less glamorous part of development: running the program repeatedly, trying to break it, finding the strange button behavior, catching the session that restarts when it should end, testing missed workouts and schedule changes, and making sure the athlete’s experience matches the logic underneath it. That debugging matters because an elegant algorithm is not enough. If the athlete cannot understand what the program wants them to do, or if the software behaves unexpectedly at the wrong moment, the underlying sophistication does not matter very much.
But the foundation is now there.
A great coach spends an enormous amount of time asking some version of the same question: given where this athlete started, what they have done, how they responded, and where we are trying to go, what does this athlete need next?
For the first time in AthleticPeak Pro, we have built a system specifically designed to keep asking that question throughout the program.
Rep Builder 10 is the first program powered by it.
It will not be the last.
Rep Builder 10 — Powered by MindLevr™ Adaptive Training Engine
Future AthleticPeak Pro training programs will be Powered by MindLevr™.
Find your next program with the program selector.
Originally published when AthleticPeak Pro was called Smolov Peak Pro.
Rep Builder 10
Choose one or more bodyweight movements, set your starting ability and ten-week goals, and train a 10-week adaptive path toward a realistic rep milestone. MindLevr™ adjusts the progression from your completed workouts rather than handing you ten fixed weeks in advance.
View the programGet new articles by email
Occasional notes on training, coaching and athletic performance. No daily noise.
SubscribeRelated articles
The Coach's Tech Stack: What You Actually Need to Train Athletes
Most coaches need three tools, not thirty. Here’s what you actually need to plan, track, and talk to your team; what to add later; and what to skip.
A Full Roster, Two Training Plans, and a Saturday Meet
Follow Coach Rivera as she balances 24 athletes, coordinates with a strength coach, and prepares for Saturday’s meet. See how AthleticPeak Pro helps her turn shared feedback, training records, and team schedules into a flexible plan for the week ahead.
One Week, Two Coaches, and a Saturday Meet
Follow Alex through a week of training, school, and a Saturday meet. See how AthleticPeak Pro helps him coordinate with two coaches, track progress, connect with teammates, and keep his family in the loop.