Cognitive Gym

Why this, why now?

Traditional learning paths worked because the work itself was the teacher. Knowledge workers got strong by doing the repetitive, unglamorous tasks: data wrangling, writing tests, generating reports. Those reps are how expertise compounds.

The problem: AI is increasingly performing tasks in areas of knowledge work that were historically reserved for humans, like software engineering, scientific research, and product design. This has resulted in an accelerating loss of job description and has taken away many of those entry-level, expertise building reps.

We have argued that the same platform change taking away the learning opportunity will also become the playground people use to acquire new expertise, as we have already started to see play out.

History repeats, just a few inches further up

For most of human history you didn't need a gym. Farming, building, hauling: daily survival was the workout. Then the Industrial Revolution moved millions of people off the land and into factories and offices, and by the back half of the 19th century the work itself had gone sedentary. The body wasn't part of the job description anymore.

Gyms as we know them trace back further than you'd expect. Frederick Ludwig Jahn opened one of the first, the Turnplatz, in 1811, when industrialization was just getting started. It took another century for the idea to really catch on: Jack LaLanne opened the first modern American health club in 1939, and in 1947 Vic and Armand Tanny launched the first health club chain, turning the gym from a niche pursuit into a place ordinary people signed up for. By the 1970s and 80s, aerobics classes, home workout tapes, and health club chains had made it a mainstream habit: equipment, classes, memberships, all aimed at people whose jobs no longer gave them a reason to move. The pattern holds up across a century of expansion: once a kind of effort stops being how you make a living, people start paying to do it anyway, on purpose, for what it does for them.

The same shift is underway in knowledge work, a couple hundred years later and a few inches further up. Factories automated away the need for physical effort as a livelihood; AI is doing something similar to cognitive effort. The tasks that used to build judgment (writing the first draft, debugging the gnarly edge case, wrangling the data by hand) are quietly disappearing from the job description, the same way heavy lifting disappeared from most jobs a century ago.

Gyms turned out to matter for more than health: moving your body and testing it against resistance is tied to feeling purposeful, not just staying fit. A cognitive gym might matter even more, for a similar reason. Physical strength was rarely what set humans apart from other animals; reasoning, planning, and building comes closer. If losing physical effort as a livelihood was enough to build an entire industry around getting it back on purpose, it's worth asking what losing cognitive effort does.

So here's a formal definition: a cognitive gym is any service or platform built to let people keep practicing the thinking their jobs no longer require, deliberately, the same role the gym has played for the body since the 1800s.

The methodology underneath

The structure comes from KnowledgeOps, a framework for turning expert judgment into repeatable, teachable process. Every phase of the build produces a concrete artifact (problem framing document, evaluation dataset, architecture spec) that carries forward into the next phase. Learning is embedded in the work itself.

Agentic Buildcamp covers our instance of the cognitive gym in detail: the program, its terminology, and how Sherpa-B fits in.

If you are evaluating fit before enrolling, start with Is This For You? for the full decision map.