Hercs: AI-powered, Sensor Drivien Personal Coach
Background
What is Hercs?
Hercs is an AI-powered, sensor-driven strength trainer that pairs its proprietary eBell smart motion sensor with an intelligent coaching app. It recognizes exercises, counts reps, monitors form, and delivers real-time feedback to reduce the risk of injury. Over time, Hercs adapts and re-optimizes personalized training plans based on fatigue, performance, and consistency—ensuring every workout is safe, effective, and motivating.
Why this project?
Hercs, originally a hardware startup, had built a robust motion sensor (eBell) capable of tracking detailed movement data. But with new AI capabilities and limited resources, the team wasn’t sure where to focus their efforts for maximum user impact. They brought me in to help rethink the product from the ground up.
This case study covers how we repositioned Hercs from a hardware-focused product for experienced athletes to a holistic AI fitness coach for beginner and intermediate users—and redesigned its core experience around safety, motivation, and progress.
Role & Team
What was my role in the team?
As the Senior Product Designer on a three-person design team, I was responsible for guiding the strategic direction while actively collaborating across all phases of the project. My contributions included:
User Research
I co-led foundational research efforts, including surveys and interviews, and synthesized findings into actionable insights that helped the team refocus on beginner to intermediate users
Product Strategy
I worked directly with the founders to define Hercs’ new product narrative—transitioning from motion tracker to AI strength coach—and ensured the team aligned on user value and business goals
MVP Scoping
I drove the scoping process by identifying high-impact use cases, aligning technical feasibility with user pain points, and helping the team converge on a focused MVP roadmap
UI/UX Redesign
I led the redesign of the core user experience, including onboarding, real-time feedback, and adaptive planning—while regularly critiquing and iterating on flows with the team to push design quality
The Challenge
Motion Tracking is cool, but not enough.
Before I joined the project, Hercs functioned primarily as a motion-tracking tool: the eBell sensor could accurately detect exercises, count reps, and record performance. While technically impressive, users often didn’t feel they were getting meaningful value—there was no feedback, no guidance, and no sense of progress.
We tracked the motions, but what value were we really delivering to the user?
2/10
Reps
15
Lbs
As the team looked to introduce AI, the question became:
Where should we apply AI to truly improve the user experience—especially for beginners and casual lifters who struggle with motivation, form, and consistency?
With limited resources and no clear direction, they brought me in to help rethink the product from the ground up:
How might we transform Hercs from a passive tracker into an intelligent, supportive fitness coach—one that delivers daily value, adapts to users’ needs, and drives real progress?
This question became the foundation for repositioning Hercs into an AI-powered training companion.
User Group
Who are we designing for?
Reframing Who Hercs Is Really For
Before we dive into the core value we can bring to the users, we took one step back to review who are we designing for. Hercs was originally designed with workout pros like experienced lifters and semi-professional athletes in mind, users who already had a strong understanding of form, training plans, and motivation. However, through user research, we quickly realized that this group didn’t need tech support as much as the team had assumed. They already had structure and our technology didn’t see added value in their daily training.
Instead, we uncovered a more promising and underserved user group:
beginner to intermediate strength trainers, people who are motivated to work out but often feel unsure, unsupported, or inconsistent when training alone.
Workout Pros
Knows what they are doing, with our resources and tech, not much we can do
Workout Bignners
Struggling to keep workout habit, need instruction and plan, yes we can help them a lot !
Market Opportunity & Growth Potential
The shift from advanced users to everyday progress-seekers unlocked a much broader market:
Original target (experienced lifters): ~10M users in the U.S.
New target (entry to mid-level fitness users using digital tools): ~60M+ users, and growing fast due to increased demand for at-home and hybrid fitness tools post-pandemic
This new group is also more likely to pay for coaching and support services than advanced users, who often rely on their own routines
By repositioning Hercs for this broader audience, we not only improved product-market fit but also opened up significant room for user acquisition, retention, and subscription growth.
Research Insights
Validate the assumptions with users
To validate direction, I led a mixed-method research sprint, including surveys and 1:1 interviews with fitness app users across different experience levels. We uncovered a key insight:
“Users often struggle to stay safe and motivated when working out on their own.”
What users really wanted was a coach who’s always there—giving guidance, tracking form, and helping them stay consistent, without pressure.
Problem 1: Lack of guidance
Injury risk from poor form & lack of guidance during solo workouts
Problem 2: Lack of motivation
Lack of motivation due to slow progress or unclear results
Problem 3: Unrealistic plans
Low retention even with existing fitness apps, often due to rigid plans that don’t adapt to real life
Design Strategy
What’s our design goal?
I proposed a strategic shift: Hercs should become a personalized AI strength coach, using the eBell sensor not just for data capture but for in-session correction and long-term training adaptation.
This meant optimizing for two outcomes:
What’s our MVP focus?
3 High-Impact Scenarios
To align limited resources with maximum value, we scoped MVP around three core use cases:
1
Personalized Plans
Set up AI-generated workouts tailored to each user’s fitness level and goals
2
Real-Time Tracking and Support
Provide instant, in-session feedback through live performance tracking and exercise guidance
3
Post-Workout Insights
Visualize progress over time and adapt future plans based on fatigue, performance, and consistency
Design
Let’s take a closer look at the launch experience
We designed Hercs as an AI-powered strength coach that combines real-time motion tracking with adaptive, in-session guidance to help everyday users train smarter, safer, and more consistently.
AI-generated Training Plans
Instead of relying on a rigid questionnaire, we prousers can now describe their fitness goals naturally through an AI-powered input field. Hercs’ GPT-based assistant interprets goals and recommends a tailored plan—offering flexibility without sacrificing structure.
Guided Path
Conversational Input
Workout session - Tracking
To keep users engaged, we designed the session view around exercise demos and real-time feedback. Users see correct form via video and receive live metrics (tempo, range, reps) captured by the eBell sensor.
Workout session - Motivation
We introduced an optional lightweight AI prompt system—using short, encouraging audio/text nudges like “Push harder” or “Keep going” to motivate users during a workout without disrupting flow.
Form Correction
When the eBell detects risky form, the AI steps in with a clear alert and offers a quick tutorial to help users correct it and avoid injury.
Post-Workout Insights
When users miss a session or feel fatigued, Hercs doesn’t punish them. Instead, the AI dynamically adjusts future intensity and recovery based on performance trends—helping users stay consistent without burning out.
Reflection
What I learned
Designing for real-time feedback with AI requires balancing clarity, confidence, and non-intrusiveness—especially in high-effort scenarios like workouts.
Hardware+AI products need to anchor user value not in “smartness,” but in concrete outcomes like progress, motivation, and safety.
What I’d do differently
Next time, I’d bring engineers into prototyping sessions earlier. Some of our most ambitious ideas around feedback visuals ran into sensor limitations, which could have been surfaced sooner with tighter design-tech collaboration.
Open to
work
Let's Chat!