There Are No Bad Students – Just Bad Matches

01 /The Concept

Didask built a platform for creating and delivering training courses – a Learning Management System, or LMS. The startup claims it built the "first LMS with educational AI."

On the course creation side, the AI does the heavy lifting. Upload your training materials – documents, slides, reference content – and the platform automatically structures them into lesson sequences, generates content, formats it, adds knowledge-check exercises, and layers in other useful elements.

Didask's chosen market is corporate training. Its customers are companies running internal employee development programs and training firms delivering corporate workshops.

Founded in France in 2017, the company ran without outside funding for years. In 2022, it added an AI engine to its standard LMS – and growth kicked in. Revenue has doubled every year since. That track record led to the company's first-ever fundraise: €10 million.

Course creation is what the AI handles first. But the more interesting capability comes after.

02 /Why It Matters

Didask noticed something telling about the LMS industry: most platforms obsess over learner engagement. They add features to keep people watching until the end of a video, moving through all the lessons, staying active on the platform. They make lessons "fun," more interactive, gamified.

Engagement goes up. But is engagement the end goal of training? Obviously not. So Didask draws a sharp line: "Engagement is good – but progress is far better."

In other words, Didask optimizes not for learning activity but for learning outcomes. The mechanism is adaptive learning.

The AI continuously and automatically personalizes the learning experience for each student based on how they're performing. Lesson difficulty, pacing, session length, feedback delivery, and the mix of practice exercises all adapt in real time. The adaptation logic draws on performance data from assessments after each lesson.

If a student is struggling, the difficulty drops and the system cycles through different exercise types to find what actually works for that person. If they're progressing well, complexity keeps increasing – until they either master the material quickly or hit a natural plateau, at which point the system settles at the most effective level.

Soon, Didask will add an AI mentor – a conversational interface learners can consult with questions and requests for additional explanations. On demand, it provides supplementary information and explanations – the obvious use.

A subtler but equally important use: The AI mentor can track what questions a learner asks, and which explanations actually help. That behavioral signal feeds back into the adaptation engine, enabling even finer-grained personalization of future material for that specific learner.

Taken to its logical conclusion, the platform effectively has no static courses – yet it has courses for everyone. This maps to the TRIZ concept of an "ideal system": one that doesn't exist as a fixed artifact, but whose function is fully realized.

Traditional LMS platforms serve the same course to every learner. Didask's courses are dynamic – each student receives a unique sequence built specifically for them, generated in real time based on how they're learning.

From this angle, the underlying knowledge a learner needs to acquire is the core product requirement. The course is just the interface through which that knowledge is delivered. It's no coincidence that Didask references "LXD" (Learning Experience Design) in its writing – a direct analog to UX (User Experience Design) in software product development.

This parallel is worth sitting with. For decades, software UIs were also static – the same interface for every user.

The emerging concept of "living interfaces" flips that: AI builds the interface in real time, adapting to how each individual user actually interacts with the product's functionality.

Fibr ([related review](/review/kak-uvelichit-konversiju-na-300)), which is working in this direction, recently published a living interfaces manifesto on its website – partly prompted by Unusual, another startup pursuing a similar platform.

What Didask is building might best be described as "living courses" – ones that reshape themselves in real time for each learner, the same way living interfaces reshape themselves for each user.

03 /Opportunities

Living interfaces and living courses are both purely AI-native concepts. Neither would have been feasible – or even conceivable at scale – before large language models. These are genuinely new territories. And building in new territory, before the obvious approaches have been tried and optimized by everyone else, is one of the clearest paths to an interesting company.

The broad direction: build platforms that "animate" – with AI – things that have historically been static. Two obvious starting points are the examples above: living courses and living interfaces.

The most compelling starting points are domains where valuable content already exists in static form – textbooks, training programs, legal reference materials – and where engagement rates with that content are chronically low. That's where animated AI stands to do the most work.