After conducting an external review of various adaptive training programs, we have observed that they all seem to operate on a similar fundamental basis. Typically, a group of running coaches has created a basic set of rules or predefined training plans that their platform follows. These rules are often based on anecdotal evidence and common practices. However, the problem lies in the simplicity of these rules, as they lack personalization and an evidence-based foundation – they are common practice. It’s worth noting that this issue is not limited to online training programs but is prevalent throughout the entire training industry, and we believe a contributing factor to the rate of Running Related Injuries being unchanged over the last 50 years.
Even platforms that claim to utilize (with great marketing) advanced AI techniques are essentially implementing a variation of the Training Stress Balance model. This model, which was developed almost 50 years ago, has known limitations and relies on laboratory testing to validate its applicability to individuals. It also assumes that any workout leads to improved fitness and that performance only improves with reduced training.
TrainAsONE sets itself apart as the only platform to employ machine learning on real-world data from all its users. This enables the platform to generate personalized training plans tailored to individual goals. Furthermore, TrainAsONE’s approach allows for constant adaptation based on each run or missed run, weather conditions, multiple races, and greater flexibility in configuration. While TrainAsONE plans are already impressive, we are continuously working on further development and have many exciting features in the pipeline. Our competitors cannot truly match these capabilities as they persist with their simplistic approach.
Consequently, our plans may sometimes deviate from the typical norm. Nevertheless, our users consistently achieve excellent results, often with less training volume. This applies to beginners as well as those competing and triumphing in extreme races.
Our approach originated from the desire to train efficiently and effectively while minimizing the risk of injury. We recognized that the majority of training practices lack an evidence-based foundation, and even those supported by scientific research are often not applicable to specific individuals or the situation at hand.
Considering the complexity of human physiology and the diverse needs and desires of individuals, training cannot be reduced to a set of simple ‘golden rules’. TrainAsONE embraces this notion, and we firmly believe that our approach is the right one. Rather than simply digitizing common practices, we strive to enhance training methodology and further the science of running.
An interesting comparison (analogy) would be recent game-playing AI systems, for example Google’s AlphaZero (for chess) and AlphaGo (for Go). Not only do these systems beat humans, but they have achieved this with novel strategies and moves that humans had never conceived (despite 1000’s of years of accumulated game study).