New AI coaching app wants to work with coaches, not replace them

Artificial intelligence is fast becoming an essential tool across many industries, and the WorldTour peloton is no exception. 

From predictive nutritional analysis and mathematical modelling, teams are turning to futuristic tech to streamline processes and deepen analysis, as highlighted in the aforelinked feature, with the ultimate aim of increasing performance. 

Separately, artificial intelligence in cycling apps is far from new. The biggest, TrainerRoad, has been actively using machine learning since 2021 and can nowadays detect a rider's FTP (Functional Threshold Power) using AI. Xert was doing similar even before that, and others like Spoked and Join are the established names among the dozens to pop up in recent years. 

Vekta, launching today, aims to do things a little differently. 

While many of those other apps try to work in place of a human coach, Vekta's aim is to provide a platform for the coach, streamlining processes so that athletes, professional and amateurs alike, can get the best of both worlds.

While it is open to self-coached athletes too, Vekta's aim is to enable AI to automate the legwork that coaches do on a daily basis, such as analysing intervals and quantifying performance. It will even build a structured workout file with little more than a text prompt.  

According to the brand, it has already been adopted by a cohort of pro teams, including Jayco AlUla, FDJ Suez, and Arkea B&B Hotels. 

Using an analysis of recent workout history, the app will build an understanding of a rider's capabilities, including their Critical Power and W' (pronounced W Prime, W' is measure of your anaerobic work capacity - essentially, the amount of energy you can produce above your Critical Power before you fatigue) while creating an ever-evolving picture of a rider's training zones.

An inbuilt chat function will enable self-coached athletes or coaches alike to unpack the numbers, and the integration of HRV data from devices such as a Whoop band, coupled with a daily wellness check will allow coaches (and self-coached athletes) to understand training readiness and an athlete's proximity to overtraining. 

Integrations with other third-party sensors are available too. Data from a Core temperature sensor can be fed in, for example, and the app can push sessions to Zwift's ZWO format to allow riders to perform prescribed workouts on their smart trainer. 

At the time of launch, the ability to push workouts to bike computers is in development, but it will at least pull in completed workouts automatically. From here, your sessions will be automatically analysed and categorised, labelled to reflect the type of session undertaken. 

Following a short demonstration of the app from Vekta's founder, it's evident that the brand is taking aim at the well-established platform TrainingPeaks, which has been around since 1999 and is immensely popular among coached and self-coached athletes alike. 

But TrainingPeaks has taken its own steps to modernise over the past year, with the acquisition of the upstart indoor cycling app IndieVelo, which it launched as TrainingPeaks Virtual. 

It's not the only development in the space, either. TrainerRoad announced a partnership with Zwift just a few months ago, which allows users to complete TrainerRoad workouts natively in the Zwift app. 

And just last week, Zwift announced an overhaul of its Companion app that saw a significant increase in fitness tracking. Notably among the updates was the inclusion of short-term and long-term training load, tracked over 7- and 42-days, the same timeframe used by TrainingPeaks' Acute Training Load (ATL) and Chronic Training Load (CTL), respectively, and how its TSS (training stress score) builds over time. 

Interestingly, though, Vekta says it will eschew the use of TSS entirely, and will instead use kilojoules burned, scaled by rider weight. 

Vekta is available as a browser-based application and a free mobile app. A free 30-day trial is available, after which a monthly fee of €19.99 / £17.99 / $21.99 will apply. Annual subscriptions are priced at €179.99 / £159.99 / $199.99.