Spot Me
Ai FITNESS APP
Preventing weightlifting injuries through real-time AI form feedback.
Overview
How might we leverage AI to make personalized coaching accessible to casual lifters?
There is no convenient, low-cost way for casual weightlifters to have their form checked. Without access to personalized guidance, most lifters have to figure out their form on their own and often don't realize something is wrong until they're already hurt.
Solution
Spot Me is an AI-powered fitness coach that provides lifters with real-time form feedback.
Real-time Form Analaysis
Leverage human pose estimation technology to analyze users' form by recording themselves on their phone.
Feedback Prevents Injuries
Provide corrective and encouraging feedback to support progress and prevent injuries.
Motivational Tracking
Encourage users to lift with correct form, using statistics to track progress and achievements over time.
problem
Most weightlifters don't realize they're lifting wrong until they get injured.
Weightlifting is one of the most popular forms of exercise and also one of the most injury-prone. In early interviews with lifters, ranging from beginners to veterans with years of experience, a clear pattern emerged: they discovered they had been performing lifts incorrectly only after sustaining an injury. No one was around to correct them.
Pain points
1
Poor form compounds with intensity
A lifter can't stop mid-rep to read a report. Any corrective guidance needs to land in the moment it can actually prevent injury.
2
Personal trainers are out of reach
One-on-one coaching was the universally desired solution, but the cost held people back. Especially for beginners who couldn't yet justify the investment.
"The biggest pain is that I don't have a trainer. How can I know I'm doing something wrong if no one's there to correct me?"
Dennis, Lifting for 10+ years
research
How are people actually building their workout routines?
I conducted six semi-structured interviews and mapped the current landscape of available tools to understand where users were finding support and where they were falling short.
market position
Youtube Videos
YouTube is where most users start because there's a tutorial for every exercise, and it costs nothing. But it leaves users to build their own program from scratch, which proved paralyzing for beginners who didn't know what they needed.
Fitness App Programs
Apps like Centr and Runna solve the planning problem by providing users with a structured program and coach-led videos to follow. The gap is that there's no feedback for the person following along to make sure they're executing the movement correctly.
Personal Trainer
Every person we interviewed said they wanted a personal trainer, not for motivation, but for real-time correction that makes training both safer and more effective. However, with the cost being at $80–150 a session, most beginners described it as aspirational rather than realistic.
Lifters misinterpret pain and struggle as progress.
The user journey mapping exercise revealed that people misinterpret pain as progress and continue pushing until injury forces them to stop working out. The map traced a recurring pattern: overexertion led to injury, and then correction.
user journey
opportunity
What if AI could correct lifters as a trainer does?
Having someone watch you lift isn't just about safety; it's how you learn and improve. A good spotter catches the moment your form breaks down, before your body does. The question was whether technology could replicate that for any casual lifter.
Real-time joint tracking
Human pose estimation maps key joints in real time, detecting the exact moment body position breaks from correct form, before damage occurs.
No new hardware required
Filming yourself at the gym is already a common practice. HPE uses the same camera, so no new hardware or behavior is required.
design process
When do users actually want feedback?
I recruited five users to shadow their existing workout routines before developing concepts. The goal was to understand the texture of a real gym session, not what users said they wanted, but when and how feedback would fit naturally into what they were already doing.
Concept 1: Real-time alerts
Bite-sized visual and audio cues triggered mid-rep the moment bad form is detected. Users can correct immediately without pausing or breaking focus.
Concept 2: Post-set analysis
Record first, review after. Users complete the set, then see an in-depth breakdown of every alert. More useful as a learning tool than an injury prevention tool.
what testing changed
Two distinct mental modes during a workout.
Real-time alerts prevent injury during the lift. Post-workout analysis builds understanding after. The final design uses both as a layered system.
During the lift
Performance mode
Laser focus. They wanted minimal interruption and only the most critical information surfaced in the moment.
After the set
Learning mode
Open and reflective. They wanted to understand what went wrong, why, and how to fix it next time.
Information
Alerts
Progress Tracking
Goal is to take time to execute each rep correctly.
Unlike cardio, where pace drives results, weightlifting rewards patience. Slowing down and focusing on each rep is exactly how casual lifters build strength safely and avoid injury.
Focused the mid-lift UI
During a set, users only need the rep name and number. Everything else was removed to reduce cognitive load.
Removed timers and progress bars
Timers added pressure without value and were removed entirely.
Final design
Real-Time Feedback on Your Form
Spot Me analyzes form in real time through the phone camera. The moment a breakdown is detected, a bite-sized visual and audio alert surfaces mid-rep, so users can correct before injury.
Easily Follow Coach-Led Videos
Every person we interviewed had tried building their own program and given up. The home screen removes that friction by putting the day's workout front and center, with coach-led tutorials for every set.
Review Your Analysis, Post Work-Out
A post-workout recap surfaces the highlights at a glance, with the option to drill into a full alert breakdown by set.
Track Your Form Progress Over-Time
Form accuracy, alert frequency, and weight carried are tracked across every session and framed as personal achievements. Watching your accuracy improve over time is what keeps casual lifters coming back.
reflections
What did I learn?
Map to reality, not ideals
The most effective design decisions came from mapping the product to how people actually lift — not how they ideally would.
Know your technology's limits
Researching human pose estimation clarified what the system could reliably detect and where it would fall short.
© Kirsten Geiger 2026. All rights reserved.