TrainerRoad presents a streamlined software solution for cyclists, built around structured workouts, offering an intuitive user experience. The platform allows users to follow a comprehensive training regimen or utilize the more adaptable TrainNow feature, which proposes sessions based on recent activity. In practice, the TrainerRoad subscription has consistently delivered effective indoor training and noticeable performance enhancements. Its AI-driven functionalities are progressing swiftly, showing considerable promise, though some experimental features are still in development.

To optimize the TrainerRoad experience, a smart trainer and, ideally, power meters on each bicycle are recommended. The platform heavily relies on power metrics, largely disregarding heart rate as a significant input. TrainerRoad is accessible across major desktop and mobile operating systems, including Apple, Microsoft, iOS, and Android. The installation is relatively light, ranging from 400-600MB depending on the device. However, loading times can be extended, sometimes taking up to a minute to activate and display the latest training information. The platform seamlessly integrates with other cycling applications, automatically syncing workouts to platforms like Zwift and ride data to Strava and Garmin Connect.

The core of TrainerRoad is its plan builder, which enables users to construct training schedules tailored to specific events or general fitness objectives. These plans can be customized for various cycling disciplines, such as road, mountain biking, and gravel, allowing for the inclusion of event specifics like distance, intensity, and priority. Events are categorized as A, B, or C, with a crucial stipulation that A-priority events must be separated by at least six weeks to ensure optimal performance. For those with less predictable schedules, the TrainNow feature offers a more flexible approach, suggesting sessions without requiring adherence to a full training plan.

After nearly a year of usage, TrainerRoad has demonstrated significant advancements in its AI capabilities, while maintaining its foundational structure and focus. The plan builder has been successfully used to prepare for diverse A-goal events, including a cross-country MTB race, a cyclocross championship, and a UCI gravel race. Building each plan is efficient and straightforward, offering ample flexibility to integrate training into weekly routines. Additional events can be easily incorporated and assigned priority levels. The platform particularly excels at adapting to interruptions, automatically adjusting upcoming workouts if sessions are missed or time off the bike is required. If a training plan needs to be entirely revised due to altered circumstances, this can also be done with ease. TrainerRoad automatically calibrates and updates your Functional Threshold Power (FTP) every 90 days with a power meter or smart trainer, eliminating the need for a ramp test. However, conducting a ramp test when fresh and before starting a plan is highly advisable to establish an accurate baseline. While FTP may not be universally applicable to all riding styles, it remains the industry standard across most training platforms.

The software has generally been utilized as a dedicated training instrument, with an ERG-mode smart trainer for indoor sessions and scheduling outdoor rides across various terrains. However, a notable limitation is its inability to effectively account for rides without power data. For instance, substantial efforts tracked solely via heart rate on bikes lacking a power meter are not recognized, hindering adaptations. Considering the wealth of personal historical ride data and TrainerRoad's AI ambitions, this presents a missed opportunity. Heart Rate Variability (HRV) is another valuable metric that could be incorporated in the future to gauge rider freshness. Integrating HRV data from wearable devices could enable TrainerRoad to further fine-tune planned sessions to reflect a user's current physical state. Another area for refinement involves addressing discrepancies between power meters. While devices may exhibit slight variations in readings, the platform currently lacks a feature to normalize this data. Although not a simple task, existing tools allow for direct comparison of power meter data, enabling users to account for known under- or over-readings by specific percentages.

TrainerRoad stands out for its adaptive training capabilities. Its AI frequently recommends adjustments based on fatigue and recent workload. Instances of disregarding these recommendations and subsequently struggling during workouts suggest the system is generally accurate. However, its accuracy can be lower at the beginning of a training block, often erring on the side of caution, indicating a need for more data to learn and adapt. The platform also appears to be better suited for shorter, more structured disciplines. It effectively managed training for XC, cyclocross, and gravel, but struggled to meaningfully integrate a 600km audax event into the plan. This observation highlights a general bias towards intensity over pure endurance. The FTP prediction feature, currently in beta, initially provided overly optimistic estimates, exceeding previous personal bests from years ago. However, these fluctuations have since stabilized, and the current predictions are more realistic. While a decrease in FTP prediction might be disheartening for some, for me, the present numbers are a more accurate reflection of achievable performance levels.

The effectiveness of TrainerRoad is evident in the tangible results achieved over the past year, including a category win at the Welsh MTB cross-country championships, achieving a 24-hour time target in the Bryan Chapman Memorial 600km audax, and a podium finish at the Welsh cyclocross championships. From a performance perspective, each training plan phase has yielded improvements, with a recent plan showing an FTP increase from 270W to 307W in just eight weeks. This marks its highest level in over a decade, despite significantly reduced training volume. In terms of cost-effectiveness, TrainerRoad is considerably more affordable than engaging a personal coach. While a skilled coach offers a level of personalization and motivation that software cannot fully replicate, the associated costs typically range from £150–£400 per month. TrainerRoad bridges this gap with its increasingly sophisticated AI, making it an attractive option for most cyclists. The monthly cost is approximately £16.60, plus any exchange rate fees. For riders who prefer a more social or immersive experience, TrainerRoad might feel somewhat limited. Combining it with a platform like Zwift could enhance the experience but would incur additional costs. In comparison, Zwift costs £17.99 per month or £179.99 annually and includes training plan features, albeit without AI integration. TrainingPeaks, a popular platform for coaches and athletes, costs $19.95 (approximately £15) per month or $135 (around £102) annually, with optional coach-created training plans that lack AI adaptive features. Overall, TrainerRoad consistently delivers excellent results and remains a valuable tool for pursuing cycling goals, such as the UCI Gravel World Championships. It demands commitment and consistency, but its structured methodology and evolving AI capabilities position it ahead of many current training platforms, even with room for further enhancement. However, for longer, endurance-focused events, there remains some uncertainty regarding whether its intensity-heavy approach provides the most comprehensive preparation. Its commitment to structured training and continuous AI development empowers athletes to achieve significant gains, fostering a culture of dedication and self-improvement in the pursuit of athletic excellence.