Race preparation is a series of decisions. What power can a rider sustain? How should that effort change across the course? Does the latest aero test change the equipment choice? Feedback from professional coaches shows how Best Bike Split (BBS) and AI can help answer those questions, from time trial pacing strategy to post-race analysis.
One coach described using the connection to prepare plans for several riders, explore race scenarios, analyze road tests and revisit past performances. BBS provides the performance modeling. AI helps coordinate the work and interpret the output. The coach decides what to test, what to trust and what the athlete should do.
Building Individual Time Trial Plans for Multiple Riders
The coach prepared individual time trial plans for several riders competing at a recent world championship. Each athlete brought different power capabilities, aerodynamic settings and equipment. Working across those plans makes it possible to consider the group while keeping each rider’s needs in view.
The useful questions are specific: which assumptions need another look, where does a rider’s profile change the demands of the course, and which plan deserves further testing? AI can help organize those comparisons while BBS calculates the modeled performances.
Running What-If Race Scenarios in Minutes
In one afternoon, the coach built more than ten race plan scenarios. This is where a connection between AI and BBS becomes useful: a first plan can lead into a sequence of questions about power targets, weather and aerodynamic assumptions.
The value is having alternatives to review before choosing a strategy. A useful approach is to change one assumption at a time, compare the resulting plans and identify which differences matter for the athlete. On that course, BBS showed that a small change in power was worth a few seconds, while the choice of bike setup was worth far more.
Turning Aero Field Tests into Race-Day Decisions
The workflow also included repeated out-and-back road tests to compare helmets and assess aerodynamic drag. These tests connect an equipment question with a race question: what did we learn, and what could it mean on this course?
BBS analysis and race modeling provide a way to explore that relationship. AI can help organize the test information and comparisons. The coach weighs the results alongside test conditions, comfort, fit and the rider’s ability to hold the position throughout the race.
Post-Race Analysis: Comparing the Plan to the Result
After the finish, the coach rebuilt historical time trial plans and compared modeled performances with official results. In one analysis, an athlete’s harder-than-modeled opening effort helped explain the gap between prediction and performance, and the race breakdown below shows what that looks like in practice.
Sending Workouts to TrainingPeaks
The workflow also included creating workouts and sending them to coached athletes’ TrainingPeaks calendars. That connection helps carry planning into the athlete’s routine. A lesson about opening pacing, for example, could become something the rider rehearses before the next event.
Case Study
Championship Time Trial (Before, During, & After)
To make this concrete, here is how one rider’s time trial at a recent championship looked before, during and after the race. We’ve kept the rider anonymous and the numbers general, but the analysis comes from the real plan and ride file.
Before the Race: Building the Pacing Plan
The coach built the rider’s plan on the official course. BBS predicted the finish time and laid out where the effort should go: a little under target on the fast opening kilometers, steady through the flat middle, and a clear lift for the rise to the finish, where the plan called for the highest power of the day.
The race plan used the rider’s time trial bike profile, with an aerodynamic drag value (CdA) the coach had entered from earlier testing. The plan also used an expected weather input for the course. Conditions on the ground turned out to be different, which matters later. Here is the plan on the course, followed by five versions of it built in BBS in a few minutes.
During the Race: Using Early Starters to Update Plans
A time trial is a staggered start. Riders leave the ramp every minute or two, and at a championship some courses are shared across categories: the elite women and men rode the same course on the same day, with the women starting hours ahead of the men. That creates a window most coaches never fully use: every rider who finishes early is a live test run for the riders still warming up.
Here is how a coach can use it.
-
Pull the first files. As soon as an early starter’s ride syncs to TrainingPeaks, BBS can analyze it in BBS Analytics. The analysis pulls in the actual weather along the course and works out the rider’s drag, so within minutes the coach knows what the day really looks like: temperature, air density and wind.
-
Update the conditions. If the morning is colder than the plan assumed, the air is denser and every rider will be slower for the same power. That is exactly what happened here: the conditions on the ground were different from the weather used in the plan. Refreshing the later riders’ plans with measured conditions keeps their targets and predicted splits honest.
-
Check the course against reality. Early riders show how the course actually rides. Are they carrying more speed through the corners than the plan expected? Is the final rise costing more time than modeled? Official time checks from the early starters add another view of where time is being won and lost.
-
Adjust strategy, not just numbers. If the wind has picked up on an exposed bridge, it may be worth shifting effort there. If the finish is decisive, a rider might hold a few watts back through the middle to hit it harder. The updated plan goes into the warm-up briefing and onto the rider’s head unit before they roll up to the start.
Caution: don’t overreact to a single file. One rider can have a bad corner or a slow change. Look for patterns across several early finishers before changing a plan.
An example prompt: “Our first three riders have finished. Pull their files from TrainingPeaks, compare the conditions and corner times to their plans, and update the plans for the riders starting in the next two hours.”
After the Race: Finding Where Time Was Won and Lost
After the finish, the rider’s ride file went straight from TrainingPeaks into BBS Analytics, and the rider had beaten the original plan. A fair comparison needs a plan built on what actually happened, so the coach recalibrated it to the ride: the power the rider actually produced, the drag BBS Analytics measured, the race-day weather and the tires actually raced. The recalibrated plan landed on the rider’s finish time almost to the second.
Then came the interesting part. The total matched, but the sections did not. The plan was noticeably too fast through the technical opening, about as much too slow across the long, fast bridge roads, and a little too fast through the technical finish. The errors cancelled out, which is exactly why it pays to look beyond the finish time.
This is where AI earns its place. The coach asked a plain question: on the long straights, how does the rider’s drag compare with the plan’s, stretch by stretch? The assistant pulled every segment of the plan and every segment of the BBS Analytics breakdown, lined them up on the same stretches of road, and compared speed, power and drag. On the bridges the rider held a CdA close to 0.17 against the plan’s roughly 0.20, and rode several km/h faster on the same power. That one finding explained most of the fast-road gap, and it took minutes rather than an evening in spreadsheets.
The technical sections told the opposite story. Even riding steadily between corners, the rider sat noticeably higher than on the bridges, and the generic corner model sized some corners too small and others too large. The biggest miss came from the course model itself: a sharp turn at the end of a long straight fell between two long course segments, so the plan treated it as a gentle bend while the rider had to brake hard and sprint back up to speed. The AI found it by comparing the plan’s segment bearings with where the rider actually braked.
Pacing told its own story. Compared with BBS’s optimized pacing, the rider went harder through the technical opening, a little easier on the fast middle, and emptied the tank on the final climb. That is a coaching conversation rather than a model error.
For the coach, that adds up to clear takeaways: calibrate the bike profile from race files, not only from testing; expect this rider to be faster than a single drag number suggests on long straights and slower through technical sections; and put the hardest corners, not the gentle ones, on the practice list.
Put back on the course map, the pattern is easy to see.
Coming Soon: Rider-Specific Cornering and Braking Models
Individualized cornering, braking and acceleration
Today, BBS models corners, braking and accelerations using general assumptions tuned to the road surface and course. That works well on open, flowing courses. On a technical course like this one, the analysis above shows it can become one of the biggest differences between prediction and result, along with how finely the course itself is segmented.
We are adding each rider’s own handling abilities to the model: how much speed they carry through corners of different shapes and surfaces, how late and how hard they brake, and how much power they spend getting back to race speed, which also shapes how a plan paces each corner exit. We are also making course segmentation corner-aware, so a sharp turn can no longer hide inside a long straight segment, and looking at how a rider’s position changes with the road, from fully tucked on the straights to sitting up through technical sections.
Every ride file run through BBS Analytics adds to that profile, so predictions sharpen the more a rider trains and races. For coaches, that means a plan built around this rider rather than an average one. It also opens new questions worth asking: how much time is at stake in a particular sequence of corners, and is it worth practicing before race day?
Why Coaches Should Connect Best Bike Split to AI
The story here is not unusual. Every coach has had a race where the result and the plan didn’t match, and most never find out exactly why. Connecting BBS to an AI assistant makes that question quick to answer. The same conversation that built the plan can pull the ride file from TrainingPeaks, run it through BBS Analytics and explain, section by section, where time was won and lost. That is minutes of work, not an evening with spreadsheets. For coaches looking for practical AI coaching tools, it turns cycling power analysis into a conversation.
It also covers the whole race cycle. Before the race, build and stress-test plans for every rider. On race day, use the early starters to sharpen the plans for everyone still warming up. After the race, turn every ride file into a better plan for the next one. BBS does the physics, AI does the legwork, and the coach makes the calls.
Getting started is simple. Pick one upcoming race, build a baseline plan, ask a few what-if questions, then bring the ride file back after the finish and ask where the time went. The goal is to turn athlete data and modeled performance into decisions the rider can use.
Connect Best Bike Split to AI Apps
Connect your account once, then talk to your assistant the way you'd talk to a coach. Best Bike Split coaches can do all of it for athletes on their roster who have granted access — read their data, build and update their plans and bikes, analyze their rides, and send workouts to their calendars.
More on Connecting Your AI App
