Strava’s new instant workout feature has got us here at The Running Algorithm wondering what the future might hold for apps like ours. While this feature wasn’t executed with the quality we’ve come to expect from the most prominent giant of our industry, it was an interesting insight into their strategic direction as an organisation. The feature is a subscriber-only training tool that automatically generates a weekly set of structured workouts for users. Each week, users are given five workouts across four different training categories with the intent of providing a ready-made, balanced training plan without manual setup. The reviews have not been positive, with DC Rainmaker describing it as “the feature, as currently implemented, is proper dumpster-fire level”; it's a clear sign on Strava’s thinking when it comes to what user’s future expectations for training software might be.
What do we think users will expect?
In 2026, users of digital training platforms will expect 100% flexibility with their training, seamless device integrations, continuous optimisation and prescription (and adjustment) that is as good as what many human coaches are able to prescribe. There will be significant consumer pushback from half baked LLM based prescription products and this will lead to a reduction in AI branding, even for platforms that make use of artificial intelligence algorithms.
Complete flexibility will mean that users are able to change any aspect of their plan such as the amount of time available per run, the number of runs in a week, when in the day these runs take place as well as other activities that aren’t running. Continuous optimisation will mean automatic plan adjustment after each activity is adjusted. This is what the market will expect within the next 12 months.
What technological and feature advancements can we expect?
The first quarter of 2026 will yield features from all apps of our nature that allow users’ training to be adjusted due to predictable and unpredictable factors. The advanced apps will make individualised adjustments and learn various aspects of users’ traits (which could be anything from typical period duration and severity all the way to individual differences in pain reporting for soft tissue injuries) and the more basic versions will be standard adjustments based on user settings (so some line of code like if pain > 8: moverun() would be run).
The execution of these features will separate user satisfaction significantly as ChatGPT often (not always) gives good advice as to what a runner should do in these situations, bad adjustment or an LLM API call won’t add value. This will lead to there being an app available at the end of 2026 which essentially prescribes and adjusts training as well as a medium to high quality human coach. Companies that cannot meet this bar will leak users to those that do.
How will the running market change?
Anyone involved in the sport of running will have felt the growth in the number of runners recently, at least in the UK and Europe. This will most likely continue, running is addictive and with the current generation of 18-24 year olds drinking less than ever and the growth in popularity of “being healthy” (amazingly, this hasn’t always been popular) the consumer market for running will continue. Running event participation surpassed 50 million runners in the US alone in 2024 and while 2025’s statistics aren’t out yet, relative to other industries our team has been involved in, the market conditions in running currently are very positive.
The trouble that consumer facing coaching applications face is that ChatGPT presents a decent, but not great, free alternative. This means that the minimum acceptable bar for what consumers will pay for has been raised from anything that beats a generic plan downloaded off the internet. Users will expect a level of personalisation that can only be reached by genuine machine learning, not just a ChatGPT API call and some linear regression that adjusts training zones. 2026 will lead to a significant increase in quality of AI coaching apps, a significant increase in AI slop and a decrease of these apps marketing their products as “AI”.
