AiRO Vayu Launch Brings Aero Testing Into the Bike Fit

AiRO has launched Vayu to bring near-instant aerodynamic predictions into bike fitting, challenging the wait between changing a riding position and evaluating it. The AiRO Vayu launch moves the company’s existing simulation platform toward interactive feedback. Its central promise is practical: explore more adjustments while the rider and fitter are still working together.

The comparison is with a familiar process: select a position, test it, inspect the result, and decide what to try next. Wind tunnels remain valuable because they measure physical behavior. Vayu changes how quickly fitters can explore candidates before deciding which deserve further attention.

Netcompany Ineos adds a professional cycling application, with AiRO confirming plans to work with the team in 2027. That relationship gives the platform a demanding setting. It also sharpens the distinction between predicting a lower-drag position and finding one a cyclist can sustain.

The AiRO Vayu launch targets the delay inside a fitting session

Vayu’s immediate significance is the shorter interval between an adjustment and aerodynamic feedback.

In its October 29 coverage, road.cc reported that AiRO’s previous calculations took about six minutes per position. Vayu predicts drag and airflow in under a second, according to the company. Its rollout to AiRO fitters is scheduled to begin in early November 2026.

The company says its model learned from more than 30,000 computational fluid dynamics simulations, which numerically estimate how air moves around objects. Those inputs include approximately 6,900 customer cases. These are company disclosures reported in the Vayu launch coverage, rather than independently reproduced findings.

The distinction between faster calculation and faster fitting matters. A fitter still needs a usable representation of the cyclist, an appropriate starting position, and an understanding of the rider’s limitations. Software speed removes a waiting period; it does not remove those responsibilities.

But waiting changes behavior. When every adjustment requires a pause, a practitioner has reasons to test fewer possibilities and favor familiar options. Interactive feedback makes exploratory questions easier to pursue during the appointment.

Consider an illustrative fitting session where a cyclist can comfortably lower their head but struggles to maintain narrower elbows. A fast model lets the fitter compare those possibilities before spending time physically reproducing every combination. The rider can then try the most plausible alternatives.

That scenario does not establish a performance improvement. It explains why response time has value beyond a technical specification. The software can help focus the conversation while the person making the change is still present.

The launch also needs a clear boundary around “real time.” Here, the phrase concerns rapid predictions within the fitting workflow. It should not be read as confirmation that Vayu continuously measures aerodynamic resistance during an outdoor ride.

A displayed drag value is a model output. Airflow graphics can make that output easier to inspect, but their visual detail does not independently establish accuracy. The useful question remains whether the recommendation leads to a better physical outcome.

The event therefore creates a concrete tension. AiRO is making aerodynamic exploration easier to access during an ordinary professional service. Fitters must decide how to incorporate those predictions without giving them more authority than the available evidence supports.

Faster feedback puts pressure on how fitters spend their time

The strongest competitive pressure falls on workflows that make aerodynamic testing a separate, occasional exercise.

A fitting appointment already involves competing priorities. The cyclist wants comfort, sustainable effort, and a position suited to their riding. Adding aerodynamic exploration must fit within that broader assessment rather than consume it.

Shorter prediction times change that allocation. Instead of treating aerodynamics as an additional appointment or a narrow final check, practitioners can investigate it alongside positional adjustments. That is the practical proposition behind AiRO’s faster feedback.

For specialist fitters, this creates an opportunity and a responsibility. More available comparisons can support more personalized recommendations. They also create more results to interpret, explain, and reject when the apparent improvement conflicts with the rider’s needs.

The pressure is not simply to purchase software. It is to explain how a recommendation was reached. A fitter relying on experience should be able to describe that reasoning, just as a model-assisted fitter should explain uncertainty.

This is where the professional team relationship becomes relevant. Cyclingnews reported that AiRO expects Netcompany Ineos to use its system alongside wind tunnel testing. The anticipated value includes extending aerodynamic work across the squad.

The 2027 team partnership remains incompletely defined, with further details expected after the initial reporting. Team adoption should therefore be understood as a planned application. It does not establish measured race gains or confirm every proposed feature.

For a performance department, the organizational appeal is straightforward. If preliminary comparisons require fewer appointments and less rider attendance, staff can explore ideas before requesting an athlete’s time. Physical testing can then focus on selected questions.

That is an analytical implication of the proposed workflow, rather than a disclosed account of the team’s operating plan. Its value depends on whether virtual comparisons reliably identify options worth pursuing.

Professional adoption also changes what customers are likely to ask. A recognizable team connection attracts attention, but ordinary riders need answers about their own bodies and equipment. They cannot assume that a professional application guarantees relevance to their circumstances.

The same caution applies to fitters choosing a platform. A sponsor announcement creates visibility. A useful service requires repeatable setup, understandable outputs, and a clear response when the software recommends something the rider cannot maintain.

The immediate pressure, then, is on the design of the fitting service. Faster predictions raise expectations about how many ideas can be explored. They also raise expectations about the quality of the explanation accompanying the final recommendation.

Instant predictions expand the search, while physical testing checks the answer

Vayu changes the search process; physical testing still addresses questions a prediction alone cannot settle.

The core mechanism is easier to understand when exploration and confirmation are separated. Exploration asks which changes look promising under the model’s assumptions. Confirmation asks whether the expected advantage appears when the rider actually adopts those changes.

Neither stage automatically replaces the other. A wind tunnel can investigate physical configurations, but the available session still requires choices about what to test. A fast model can help propose those choices, provided its limitations remain visible.

The AiRO Vayu launch is significant because it moves more of that exploration into an interactive setting. The potential benefit comes from comparing alternatives efficiently. Raw calculation speed matters because it changes the number and sequence of decisions people can reasonably make.

For example, a fitter might find that a position looks favorable only when the rider maintains a particular head angle. That result should prompt a practical discussion. Can the cyclist see adequately, remain relaxed, and reproduce the posture without constant prompting?

A model can help identify that dependency without resolving the human question. The rider’s ability to maintain the position becomes part of the evaluation. An apparently smaller aerodynamic gain can be more useful if the cyclist can reproduce it consistently.

This is also why aerodynamic optimization should not become a competition to display the lowest number. A numerical minimum is a result within a defined calculation. A riding recommendation must account for the context in which someone will use it.

Alternative technologies approach the same problem from another direction. Body Rocket developed an on-bike system using force sensors at the rider’s contact points. A peer-reviewed study evaluated its drag measurements in a wind tunnel and indoor velodrome.

That on-bike drag study reported encouraging validity and reliability results within those settings. Its authors also identified outdoor validation as further work. The distinction illustrates why evidence from controlled conditions needs a clearly stated scope.

This comparison does not establish that either approach is universally superior. Model-based exploration offers flexibility when considering hypothetical positions. Instrumented testing investigates a physical rider and setup, while bringing its own requirements and measurement constraints.

Commercial execution presents a separate challenge. In June 2026, Body Rocket’s founder told Cycling Weekly that the business had exhausted its funding. The company was selling assets and moving toward liquidation, according to the aero sensor business account.

That history does not predict AiRO’s future. It shows why technical evidence and a sustainable customer service deserve separate evaluation. A promising method still needs dependable delivery, support, and a workable place in customers’ routines.

For Vayu, the clearest position is therefore as an exploration tool within a broader testing process. Its success should be judged by the quality of subsequent decisions. An expanding list of virtual comparisons is useful only when it helps people choose better physical experiments.

The accuracy claims require more context than a headline percentage

Choosing the faster option in a test set is useful evidence, but it does not establish universal accuracy.

AiRO says Vayu selected the faster position in 97 of 101 comparisons. It also reports selecting the faster helmet in 72 of 74 swaps. Those figures describe company-reported testing and should retain that attribution.

The first question is what each comparison used as its reference. Agreement with a separate simulation answers a different question from agreement with physical measurements. Readers need that distinction before interpreting a success rate as evidence of real-world performance.

The next question concerns the size of the differences. Identifying a large advantage is different from distinguishing alternatives whose results are nearly equal. A useful evaluation should make those cases visible rather than combine them into a single reassuring number.

There is also a distinction between ranking and magnitude. A model can correctly identify which option has lower drag while overstating the improvement. That matters when a rider is deciding whether a small expected benefit justifies an uncomfortable adjustment.

Training-set size cannot resolve those questions by itself. More examples provide more material for learning, but the distribution of those examples matters. Buyers need to know how closely the evaluation resembles the riders and configurations they intend to assess.

Academic research supports taking the underlying simulation choices seriously. A 2024 study by Thijs van Druenen and Bert Blocken examined how computational settings affect cyclist drag predictions. It found meaningful sensitivity to choices about the numerical representation and airflow modeling.

Their cycling simulation research emphasizes transparent reporting and comparisons with experimental measurements. It does not independently validate Vayu. Its relevance is methodological: the quality of the simulation foundation matters when evaluating a model trained from simulation outputs.

That creates a layered validation task. The rider representation must be suitable, the underlying calculations must be credible, and the predictive model must reproduce useful relationships. Finally, the proposed adjustment must translate into a physical benefit.

A weakness at any stage can affect the recommendation. Accurate prediction of an unsuitable body representation still leaves a practical problem. Likewise, accurate replication of a simulation does not automatically establish that the simulation describes every riding condition.

AiRO’s earlier published benchmarking provides background, but it should not be conflated with a comprehensive evaluation of the new model. Validation has a date, a product version, and a defined set of conditions. Those details should accompany claims as the platform changes.

For customers, the productive stance is neither automatic acceptance nor blanket dismissal. Ask what the evidence supports, where uncertainty remains, and how recommendations are checked. That approach gives a useful prediction room to contribute without turning it into a guarantee.

A digital rider still needs a human interpretation

The practical limitation is the gap between a modeled posture and a person riding under changing conditions.

A digital twin is a computer representation used to evaluate an individual’s configuration. Its usefulness depends on how well it represents the details relevant to the question. A convincing visual resemblance is helpful, but it is not the entire requirement.

AiRO has previously documented limitations involving body representation, equipment, and simulated conditions. Its published material describes restrictions around unusual body shapes, clothing detail, and aspects of helmet modeling. Those disclosures concern the documented platform and need confirmation against the version a fitter actually uses.

The company’s documented modeling limits are valuable because they make the boundaries discussable. They should not be treated as proof that every restriction remains unchanged in Vayu. Equally, a faster model should not be assumed to eliminate them.

Representation deserves particular attention when a recommendation concerns a small positional adjustment. If the digital body does not closely match the rider in relevant areas, apparent precision can become misleading. The appropriate response is to inspect suitability before interpreting fine differences.

A person also changes during a ride. Fatigue, breathing, attention, and the need to look ahead affect posture. A fixed representation can help compare candidates while leaving open how consistently a cyclist will occupy them.

This makes the fitter’s role more consequential as predictions become easier to obtain. The professional must translate a virtual suggestion into a position the person can reproduce. They also need to recognize when an attractive result rests on an unrealistic assumption.

An illustrative helmet comparison shows the problem. A modeled option might look favorable with the head held in a particular orientation. The recommendation becomes less convincing if that orientation disappears whenever the cyclist looks ahead naturally.

Physical fit adds another layer. A helmet model that looks aerodynamically promising still needs to fit the individual appropriately. A virtual ranking cannot substitute for checking the actual equipment on the rider.

Similar reasoning applies to posture. Narrower arms or a lower torso should remain hypotheses until the cyclist tries them. The relevant outcome includes sustainable effort and control, rather than aerodynamic appearance alone.

There is a recordkeeping implication for practitioners, too. A useful recommendation should preserve the starting position, the proposed change, the assumptions, and the reason it was accepted. Otherwise, later testing has little context for explaining a different result.

None of these considerations diminishes the value of faster exploration. They define where that speed belongs. Vayu can make it easier to identify questions, while the professional remains responsible for deciding which answers are applicable to the person present.

Watch the rollout, independent comparisons, and the team’s actual use

The next evidence should concern usable delivery, reproducible recommendations, and a clearly defined professional application.

The first signal is the early-November fitter rollout. The announcement establishes an intended availability window. What matters next is whether practitioners can use the released system during normal appointments, with a clear account of what is included.

That means separating interactive position evaluation from more ambitious automated searches. Launch demonstrations, available functions, and future capabilities should have distinct descriptions. Otherwise, customers risk evaluating the product against features that have not yet reached them.

A completed rollout would strengthen the case that Vayu has moved beyond a demonstration into an operating service. Continued ambiguity about access or included functions would weaken that case. Availability should be observable without requiring readers to infer it from promotional language.

The second signal is an independent comparison using the released model. The most informative test would describe the rider representation, alternative configurations, reference measurements, and repeatability. It should report disagreements as clearly as successful predictions.

Such evidence would help distinguish several possible outcomes. Vayu might be useful for identifying broad improvements while remaining less reliable for close comparisons. Alternatively, testing might show consistent ranking across a well-defined range of changes.

Either outcome would provide more actionable information than a general accuracy claim. A system with clearly understood boundaries can still be valuable. A system whose boundaries remain unclear is harder to incorporate into a service responsibly.

The third signal is the detailed Netcompany Ineos agreement. Readers should look for the actual scope of use, the role of physical testing, and the responsibilities assigned to the platform. A team logo alone does not answer those questions.

Evidence that virtual recommendations guide subsequent physical comparisons would strengthen the workflow argument. Public discussion of rejected recommendations would also be informative. It would reveal how the team distinguishes useful exploration from changes that do not survive practical assessment.

The AiRO Vayu launch therefore deserves attention as a change in decision timing. Aerodynamic ideas that once required waiting can become part of an active fitting conversation. The important outcome is whether that conversation produces recommendations people can use and verify.

Before booking a model-assisted fitting, ask the practitioner to explain what the software evaluates and how the final position will be checked. Request a record of the starting setup, proposed changes, and expected benefits. Treat an appealing number as a reason to investigate a position further. If you follow the professional partnership, look for descriptions of how predictions influence testing rather than assuming sponsorship establishes performance. The useful question for the next fitting or team update is concrete: which recommendation survived physical evaluation, and what evidence made it worth adopting? That is the standard against which the AiRO Vayu launch should be judged.