How to Use Free AI Tools to Analyse iRacing Telemetry (and Get Faster)

Telemetry is one of the most powerful tools available to sim racers, but for many of us it's also one of the most intimidating.

We've all heard the advice: "Check your telemetry." The problem is that once we open a telemetry graph, we're often greeted by a wall of data that looks more like an engineering report than something designed to help us drive faster.

Brake traces, throttle inputs, steering angles, speed graphs, yaw rates, lateral G-forces—the information is all there, but turning that information into actionable improvements isn't always straightforward.

A common mistake is assuming that telemetry will immediately show us the answer. In reality, telemetry usually shows us what happened, but not necessarily why it happened.

For example, we might see that a faster driver carries more speed through a corner than we do. That's useful information, but it doesn't automatically explain whether we're braking too early, turning in too aggressively, applying throttle too late, or simply taking the wrong line.

This is where many sim racers become frustrated. We know we're losing time, but identifying the specific driving technique causing the problem can take a lot of experience and interpretation.

Information Overload

Another challenge is the sheer amount of data available.

Modern telemetry tools can capture hundreds of data points every second. That's fantastic for professional teams and experienced analysts, but for the average sim racer it's easy to become overwhelmed.

Instead of focusing on one or two meaningful improvements, we can find ourselves jumping between graphs, comparing numbers and chasing tiny differences that have little impact on overall lap time.

Not Everyone Is an Engineer

Most of us got into sim racing because we enjoy driving, competing and improving our racecraft - not because we wanted to become data analysts.

Traditional telemetry analysis often assumes a level of technical knowledge that many hobby racers simply don't have. Terms like trail braking, steering rate, throttle modulation and corner phase analysis make sense to experienced drivers and coaches, but can be confusing for newcomers.

Turning Data Into Action

Ultimately, the biggest challenge isn't gathering telemetry data - it's understanding what to do with it.
Knowing that we're losing two tenths through a corner is useful. Knowing how to gain those two tenths is what actually makes us faster.

This is where AI can be surprisingly helpful. By combining Garage 61 with tools like ChatGPT, Copilot or Gemini, we can translate complex telemetry graphs into plain English explanations and practical driving advice that is much easier to understand and apply on track.

What AI Can and Can't Do

At this point, it's worth setting some expectations.

Our Workflow: Garage 61 + AI

Now that we understand where AI can help - and where it falls short - let's walk through the workflow we use to analyse telemetry and identify opportunities for lap time improvement.

Step 1 - Set Your Baseline

The first thing you'll need to do is set your baseline lap time. Make sure your Garage 61 agent is running and jump into a practice session (or do an offline test drive) in the car and track combination you're looking to work on.

Resist the temptation to do a single hot lap and call it a day. Put together a handful of representative laps and wait until your times start to plateau. This will give you a much more realistic baseline to compare against.

Step 2 - Find a Faster Driver/Lap Time

Log in to Garage 61 via web browser and do a search for lap times on the same car and track.

Spend a few minutes finding a lap that closely matches your conditions and setup (track temperature, air temperature, humidity, open/fixed setup).

One of the biggest mistakes sim racers make is comparing themselves to world-record "alien" laps. While this can be useful later, it often creates unrealistic expectations. A more effective approach is to target drivers a little faster than us and gradually close the gap. Initially shooting for 1-2 seconds faster than your baseline is a good place to start.

If you want to get more specific on what your target lap time should be in order to be competitive, a great place to check is iracingstats.net. From here you can look up any current or previous week for each series and check out the Expected Lap Times for your iRating (see screenshot below)

Based on this, a target lap time in the low to mid 1:24's would be a good target for a competitive race pace.

Step 3 - Create the Analysis in Garage 61

Create the Analysis within Garage 61 and ensure your baseline and faster (target) lap times have been added. At this point we're not trying to interpret the data ourselves - we're simply preparing the information that we'll later feed into AI.

The Garage 61 analysis provides a wealth of telemetry data to work with, and by default will show you the following graphs;

• Track Map

• Line Distance (ie. left/right distances on physical track location at any given point)

• Time Delta

• Speed

• Throttle

• Brake

• Gear

• RPM

• Steering Wheel Angle

Here's an example analysis below;

Step 4 - Take Screenshot(s)

Once you've got your analysis, you can take a single screenshot of the entire page (like the one pictured above), or if you want to focus on a particular corner, sector or other section of the track, you can zoom in on it and take more screenshots.

These screenshots form the foundation of our AI analysis. Rather than manually interpreting every graph ourselves, we're going to use AI to identify patterns and explain what the telemetry is telling us.

Step 5 - Let AI Interpret the Data

Now that we have our analysis graphs, you can go ahead and upload it into the AI engine of your choice. We'd encourage you to try any and all free ones that are available to see which one works best for you.

The most popular free AI chat bots are Microsoft Copilot, Google Gemini and OpenAI's ChatGPT.

You will need to provide a prompt when uploading the screenshot to tell the AI engine what you want to get out of it. Effective prompt generation is an art within itself, so the more detailed you are with it, the better the result will usually be.

Tip: Save this prompt as a Copilot Agent, ChatGPT Custom GPT, or Gemini Gem so you can reuse it every time you analyse telemetry. Once you've refined a prompt that works for you, there's no need to rewrite it from scratch for every session.

Example Prompt

Here's a prompt that we've been using which should provide a good starting point. You can tweak this based on your own requirements, or based on the responses that the AI engine does (or doesn't) give you.

It's quite long, but as previously mentioned, more detailed prompts tend to give better and more accurate responses. You can strip it back, include even more detail or really just drill into a particular area of your driving.

And if you need help developing your prompt, you can ask the AI engine to make one for you based on what you're trying to achieve.

AI will often give us a long list of recommendations. The temptation is to try and fix everything at once, but that usually creates more problems than it solves.

Instead, focus on the one or two areas responsible for the biggest time losses and head back out on track. Run a few more laps, generate another comparison and repeat the process.

The goal isn't to create the perfect lap in a single session. The goal is to continuously close the gap to the target lap.

If the responses you're getting from AI are too broad, adjust your prompt to focus on a specific corner, section of track or driving technique. The more focused your question, the more useful the answer will usually be.

Remember, the goal isn't to become a telemetry expert overnight. The goal is to use the tools available to us to identify weaknesses, practice deliberately and gradually close the gap to faster drivers.

By combining Garage 61 with free AI tools like Copilot, Gemini and ChatGPT, we can spend less time staring at graphs and more time doing what really matters - driving faster.

Real Example: Mercedes AMG GT3 at Mount Panorama

Mount Panorama (Bathurst) is one of the most challenging and unforgiving circuits in iRacing. When it appeared on the schedule for the Asia-Pacific GT3 Series, we found ourselves nearly a second off the pace of the faster drivers. Rather than guessing where the time was being lost, we uploaded a Garage 61 analysis screenshot to Microsoft Copilot and asked it to review the telemetry.

The response was surprisingly detailed. In fact, it went on for several pages. To keep things manageable, we've included the executive summary and the highest-priority improvement area identified by Copilot.

Actual Response from Microsoft Copilot

Mount Panorama AMG GT3 Lap Comparison
Red (baseline lap): 2:03.100
Blue (target lap): 2:02.167
Delta: -0.933s

From the telemetry, almost the entire loss occurs from Griffins Bend through Skyline/The Esses (roughly 2.5 km – 5.5 km). You are essentially equal down Conrod and only a small amount is lost in the final sector.

The key clue is the delta trace:

• Flat until ~2 km.
• Loses steadily from 2.5–4.5 km.
• Ends at about -1.0 s.
• Very little recovery afterwards.

Priority 1: Skyline → Esses → Dipper
Estimated loss: 0.45–0.60s

What I'm seeing
Your steering trace becomes noticeably busier than the target lap.
The line-distance graph also shows you travelling a longer path through the mountain.
That normally means:

• Turning too early.
• Extra steering corrections.
• Car not rotated before commitment.
• Scrubbing front tyres.

Why
The AMG likes being settled before cresting Skyline.
If you arrive with residual brake pressure and excess steering angle:

• Front washes.
• You add more lock.
• Distance travelled increases.
• Speed falls all the way to The Dipper.

Why
The car is being guided rather than flowing.
Every correction costs speed.
Even 2–3 km/h lost through Skyline becomes 8–10 km/h by Forrest's Elbow.

Exit
Your issue
Throttle hesitation.
The throttle traces are close, but Federico reaches full commitment earlier.

Driving change
Focus on:

1. Straighter arrival at Skyline.
2. One steering input.
3. Allow car to fall to apex.
4. Get to maintenance throttle earlier.

The interesting thing wasn't necessarily whether Copilot was 100% correct on every detail. The value came from how quickly it identified a clear area of focus.

Rather than telling us to improve everywhere around the circuit, it highlighted the Skyline to Dipper section as the biggest contributor to the lap time deficit and estimated that over half of the lost time was occurring in that sequence alone.

It also translated a collection of telemetry graphs into actionable driving advice:

• Arrive at Skyline with the car more settled.
• Reduce unnecessary steering corrections.
• Focus on carrying speed through the mountain section.
• Commit to throttle earlier on exit.

You can see how this level of information can get overwhelming pretty quickly, which is why it's important to focus on one or two changes at a time rather than trying to fix everything at once. We also found the short and succinct "coaching cues" really beneficial to help drive home the focus points.

AI is at its most useful when it helps us identify the biggest opportunities for improvement. From there, it's up to us to get back on track, test the advice and decide whether it actually makes us faster.

Final Verdict