The Room, Rebuilt.

Learning built machinery of its own. For generations we gave it to institutions, which bundled the teacher with a building, a brand, a syllabus and a fee. Too often, the person at the front of the class had never done the thing being taught. The internet promised to unbundle all of it, and in one sense it did, because information became free. Yet in the final year of a study of free Harvard and MIT courses, about three in every hundred participants finished.⁵ The catalogues that followed grew larger and cheaper, and this year two of the best known merged.⁶ Inside them, the share of revenue paid to the people who actually teach kept shrinking.⁷ When an institution is paid for attention and volume, the person who actually knows something becomes its cheapest input.

News went through this break first, most visibly in America. For most of the last century, readers trusted a writer but bought a newspaper. When the advertising that paid for newspapers moved elsewhere, more than half of American newspaper newsroom jobs disappeared.⁸ Two founders then argued that readers would pay the writers they trusted directly, if someone built the rails, and they called the company Substack.⁹ Millions of readers now pay writers that way.¹⁰ Writers kept writing, left the institution, and their readers followed them.

Learning has reached the same point, with one difference, and that difference is the reason we exist. A reader can judge a newsletter in two minutes. A learner who pays for a cohort is betting months of evenings and a serious sum on a stranger’s promise about their future, whether that is a job in Munich, a rental that pays its own rent, or an AI product with their name on it. A bet of that size needs someone to vouch for the teacher and someone to stay with the student, as well as payment rails. The marketplaces that last, two investors once argued, are the ones that take on the hard work of creating trust themselves.¹¹ On Maven, an American cohort platform, nearly nine in ten learners were still active in the second week, by the platform’s own count.¹² When the same platform studied what made its courses succeed, it found that expertise mattered and audience size did not.¹³

This is why the next step has to come from Topmate. Nearly thirty lakh people have a profile on Topmate, and more than three lakh of them offer what they know. For years we let each of them build a room of their own. Over the past year, quietly, we did something different with a handful of them. We sold their cohorts, ran their communities, and watched what happened to their students. Nine of them have met the standard we are now willing to put our name to. Each has done in their own life the thing they teach. Between them, they have built AI systems, shipped AI products, run a portfolio of rental homes, moved to Germany, to Dubai and across Europe, and built products at Microsoft and BharatPe. Today, for the first time, those cohorts carry Topmate’s name.

Magnor is the room, rebuilt. The practitioner teaches live, to a small group, and Topmate does everything else. We speak with you before you join and tell you plainly whether the cohort fits. We give your cohort a community manager who knows your name and notices when you go quiet. We hold you to the work, and we stay through what comes after the last session, which is where a life actually changes. Nobody can promise the outcome, and we will not. What we can promise is that you will not be left alone on the way to it. Our teachers pay us only when we sell a seat, so our income depends on every cohort being worth what it costs.