Level
Fitness progress, redefined.
Type:
Passion Project
Category:
Health & Fitness
Duration:
Oct ‘25 - Ongoing
You lift. You sweat. You skip the trainer because rent is real. Progress should be simple, so you watch the scale. It climbs.
But the mirror says your shoulders are fuller. So you try a body-composition machine. A printout arrives- fat, lean, water, phase angle, acronyms that read like a lab report. Next month, same jargon and numbers that don’t line up.
Are you getting stronger or just heavier?
Are you winning or wasting time?
You don’t need a lecture. You need a path.
Something that makes sense of the data and turns it into tomorrow’s plan.
how this gap found me
Over time I learned that consistency is not the key to the gym. After multiple injuries stalled my progress, I saw that ego lifting and impulsive decisions were the real culprits. So I locked into the same routine week after week. It produced some results, but not enough to justify the hours. Consistency alone was not enough. Progression mattered. And progression in health is personal. Many of us still use body weight as the scorecard, which misses the point. Body weight is a blunt tool. Body composition is the better lens because it separates fat, muscle, bone, and water. It shows whether you are gaining the right tissue and losing the right tissue, even when the scale barely moves.
The catch is that body composition reports often sit in a folder and gather dust. Without basic knowledge of kinesiology and wellness, most people cannot decode the metrics or translate them into training choices. Even fewer have a clear way to compare reports over time. Protocols vary, hydration swings the numbers, and the dashboards are not built for everyday athletes. The result is data without direction.
Before I try to solve for this apparent gap, I need to admit my bias. This is my lived experience, not a universal truth. That is why my next step is to observe the existing space, seek validation, test these assumptions with others, and learn what actually helps people turn numbers into progress.
scoping the landscape
Before designing anything, I needed to sanity-check the landscape. Not “are there fitness apps?” - obviously yes. The real question was narrower- if a gym-goer gets a body composition report (usually BIA), can any product reliably:
ingest the report (ideally from different machines),
explain what changed and why, and
turn that into next-week training decisions (without a coach)?
Scanner Ecosystems (hardware-first platforms)
These products do address body composition insights, but usually only inside their own ecosystem.
InBody (Popular)
InBody provides an app where users can review “results, graphs, and interpretations,” but access can vary by facility/model.
InBody’s own documentation positions the result sheet as powerful, but it still expects a level of interpretation skill (or professional guidance) to unlock value.
Evolt 360 / Evolt Active
Evolt markets “immediate insights” and “instant feedback,” plus progress graphs and even nutrition components, again, tied to Evolt scans.
Evolt also pushes a “dashboard” for insights, reinforcing that this is a platform play around their scanner.
seca mBCA / myAnalytics
seca’s myAnalytics positions itself as “success made visible” with mobile access to results and progress tracking.
These solutions are real, but they’re device-anchored. They don’t solve the broader user reality where someone might scan at different gyms / different machines, have printouts or PDFs sitting in WhatsApp/email, and need training direction, not just more charts.
Smart scales + companion apps (consumer BIA at home)
These apps are good at collecting numbers, and sometimes visualizing trends, but they generally don’t translate scans into structured training decisions.
Examples
FITINDEX emphasizes tracking multiple metrics + “analysis charts and reports.”
Withings “Body Scan” pitches richer body composition tracking and personalized biomarker insights, but it’s still fundamentally a device + dashboard model.
Great for continuous tracking, weak on “tell me what to do in the gym tomorrow.” Also: home BIA does not align with the gym report workflow we’re targeting.
DEXA/body scan companions (report viewers + trend dashboards)
These are closer to the “report ingestion” angle, but many are either tied to DEXA clinics, or focused on visualization over action.
Examples
“Body Insight” positions itself as a companion for DEXA reports: view scan reports, track metrics over time, book appointments.
Even when report access exists, the output is usually tracking and dashboards, not a training plan that adapts to the body data.
AI body scanning apps (phone-based composition estimation)
These products try to bypass machines altogether (scan with a phone), and some claim personalized recommendations.
Examples
“Spren” markets “insightful analysis and personalized recommendations” from a scanner-style experience.
“Visualize AI / Visualize Me” markets smartphone-based body composition analysis as clinically validated/patented.
These are solving a different problem: “how do we measure body comp without equipment?” Our gap is: given the report users already have, turn it into a plan.
TL;DR
We’ve solved “getting the numbers.”
We haven’t solved “what to do next.”
Body composition data exists across devices, printouts, and PDFs, but there’s no neutral layer that processes reports, explains meaningful change, filters noise (hydration, protocol), and translates it into a clear training path.
key problems
Confusing Body Data
Numbers come in- fat, muscle, water, but users have no idea what any of it means or whether it’s good, bad, or noise.
Training Without Guidance
Most people lift alone. No coach, no feedback loop, just guesswork and hope.
Inconsistent Reports
Body-scan results shift based on hydration, timing, or machine quirks, making progress feel unreliable.
Chaotic Planning
Users bounce between routines, reels, and random advice, ending up with workouts that lack structure or continuity.
Fragmented Fitness Tools
Tracking, training, and body data live in separate apps, nothing works together to tell a clear story of progress.
assumptions, exposed
1
assumptions
what research showed
1
No solution solves this cleanly
Tools exist, but fragmented and inconsistent
2
Users without trainers lack direction
Only ~11% use trainers; majority self-guided
3
People struggle with interpreting BIA
Health numeracy research confirms this
4
Users regularly take BC tests
4/6 interviewees do at varying frequency
5
Workout planning is disorganized
Interview quotes confirm chaos
6
People crave fitness improvement
Market + trend data confirm demand
note//
Secondary research showed that body composition reports reflect both training and nutrition. I had initially scoped out diet features to keep the MVP lean; I then revised the approach to include minimal, linked nutrition tracking so BIA stays meaningful without bloating the product. It’s a concrete example of evidence-driven iteration: learn fast, adjust fast, and keep the core experience sharp.
mvp features
Numbers come in- fat, muscle, water, but users have no idea what any of it means or whether it’s good, bad, or noise.
Quick and easy BC report ingestion.
Provide actionable insights after ingesting BC reports.
Access to easy-to-comprehend body metric trends.
Workout routine creation, with integrated tips and modifications after BIA logging.
Agentic workflows that solve for gaps in gym routines.
task flows
Get Oriented
Intent
Help users enter the system with confidence, not effort — capturing only what’s necessary to personalize progress from day one.
Entry
Value proposition
Sign in
Privacy assurance
Goals & training context
Body basics
Personalization
Home
Turn Raw Body Data Into Usable Signal
Intent
Transform a complex BIA report into structured, reliable input the system can reason with.
Home
Record BIA Report
Capture / Upload
Accuracy guidance
Scan & process
Confirmation
Insights surfaced
Understand Progress Without Interpreting Charts
Intent
Let users validate progress over time without requiring medical or fitness literacy.
Home
Body Metrics snapshot
View all metrics
Select metric
Trend over time
Insight summary
Context & ideal range
Translate Insights Into Training Decisions
Intent
Convert body composition insights into clear, actionable changes to weekly training.
Insight surfaced
Update Weekly Workout
Select training days
Review suggested changes
Accept / Reject / Ask Assistant
Confirm updates
the continuous loop
At its core, Level is built around a single reinforcing loop:
Measurement: captures the state of the body
Interpretation: turns numbers into understanding
Adjustment: reshapes the plan
Training: generates new signal
Each pass through the loop increases clarity, reduces guesswork, and compounds progress.
This is not a tracking app. It’s a decision system that learns with the user.
research focus 1/2
parsing body composition reports
Problem
Body composition reports arrive in inconsistent formats:
Printed slips from gym machines
PDFs from clinics
App screenshots shared by users
Each varies in layout, labels, units, and visual hierarchy, making direct data extraction unreliable.
Research Goal
Design a parsing approach that works across formats, brands, and input quality, without forcing users into rigid upload rules.
Key Findings & Approach
Product Implication
Parsing is treated as a best-effort interpretation layer, not a truth engine. The system is built to improve with usage—not break on edge cases.
research focus 2/2
homogenizing body composition data across reports
Problem
Even when data is successfully extracted:
Same metric ≠ same meaning
Brands rename, recompute, or partially expose measurements
Users compare apples to oranges without knowing it
Research Goal
Create a canonical body data model that normalizes meaning, not just names.
Key Findings & Approach
1. Canonical Fields as the Source of Truth
Define a fixed internal schema (e.g., body_fat_percentage, lean_mass_kg)
Canonical fields are:
Brand-agnostic
Unit-normalized
Explicitly defined (what it is + what it is not)
Reports never store raw brand labels as primary data.
4. Explicit Gaps > Fake Precision
If a report lacks a canonical field:
The system records absence, not approximation
Insights adapt to available data instead of hallucinating completeness
Product Implication
Homogenization enables longitudinal insight, not just report summaries. Users can track progress even when switching gyms, machines, or formats.
tl;dr
Parsing focuses on template detection + regex sets + confidence scoring, not brittle OCR hacks
Homogenization relies on canonical fields and brand-to-canonical mappings to preserve meaning
The app treats body data as an evolving signal, not static truth
Result: progress tracking that survives messy inputs and real-world inconsistency
note//
Skipped on traditional lo-fi wireframes for this project. The complexity of the system - data density, temporal context, and decision timing, meant that real problems only surfaced in high fidelity. By designing directly in detail and iterating quickly, I was able to validate hierarchy, clarity, and flow under realistic conditions. This approach reduced iteration time and kept the interface tightly aligned with the product’s core loop.
design system
Colour System
PRIMARY BRAND COLOUR
Performance Green
#71CB95
Brand highlights, Active states, Key emphasis moments, Select surface accents
BG & SURFACES
Deep Charcoal
#121212
Primary canvas. Sets the dark, performance-focused foundation.
Deep Graphite
#202020
Used for main cards and containers that sit closest to the user’s focus.
Charcoal Grey
#2D2D2D
Used for nested surfaces and layered depth.
TYPEOGRAPHY AND ICON
Soft Stone
#D4D4D4
High-contrast neutral used for primary reading content and key actions.
Muted Slate
#A4A4A4
Reduced emphasis for supporting information, metadata, and secondary labels.
Muted Grey
#878787
Used where visual hierarchy requires de-prioritisation without loss of legibility.
The catch is that body composition reports often sit in a folder and gather dust. Without basic knowledge of kinesiology and wellness, most people cannot decode the metrics or translate them into training choices. Even fewer have a clear way to compare reports over time. Protocols vary, hydration swings the numbers, and the dashboards are not built for everyday athletes. The result is data without direction.
Type System
ui concepts
Turn raw body data into usable signals
This flow was designed to collapse a traditionally complex and error-prone process into a moment of clarity. Body composition reports are dense, inconsistent, and highly sensitive to context such as machine type, timing, and hydration. Without system support, users would be forced to manually identify the device used, input conditions, and interpret unfamiliar tables, often leading to inaccurate conclusions or disengagement. By enforcing dependable parsing and recognition logic, the system allows users to move directly from a photo or document to validated insights in under 20 seconds, removing manual inputs that could compromise accuracy.
The output prioritizes understanding over data exposure. Instead of surfacing raw metrics, the system leads with workout and diet insights derived from trends developing over time, followed by clear metric context showing current, previous, and target values. An interactive trend view anchors progress across weeks, turning isolated reports into a continuous narrative. This level of comprehension is not achievable through static PDFs alone and addresses the core need for a tool that can track, interpret, and explain body composition data in a way users can actually understand and act on.
Understand progress without interpreting charts
This flow focuses on helping users understand their body metrics without requiring them to interpret dense charts or medical-style data. The primary challenge was balancing clarity with depth, conveying meaningful progress while avoiding visual overload. At the parent level, the interface surfaces only priority information for each metric: the current value, the previous value, and the target. This creates an immediate sense of direction without forcing users to scan graphs or tables, allowing progress to be understood at a glance.
When users choose to explore further, the flow transitions into metric-specific views where information is progressively revealed based on importance. Data is segmented into trends, insights, and contextual ranges, with visualizations used to carry meaning wherever possible to reduce text heaviness. Graphs are lightweight and interactive, designed to support comprehension rather than analysis, while written insights translate patterns into plain language. This structure allows users to build an intuitive understanding of their progress over time, maintaining clarity as complexity increases without ever cluttering the experience.
Translate insights into training decisions
This was the most interesting flow to design, largely because the interaction model I had in mind doesn’t commonly exist in mobile fitness products. The challenge was not generating workout recommendations, but deciding how users should engage with an agentic system making those suggestions. Drawing inspiration from permission-based interaction patterns I experienced while building a web app on Cursor, the flow presents system-generated workout changes as explicit, reviewable actions. Users can accept or reject individual suggestions, preserving agency while still benefiting from data-driven guidance rooted in body composition progressions.
When suggestions are rejected, the flow intentionally opens into a conversational, generative interface. This sequencing encourages users to first engage with the system’s reasoning before requesting custom changes, creating a feedback loop that captures preference signals and constraints. Over time, these interactions inform the agentic system, allowing it to refine future workout suggestions after each body composition update. The result is a collaborative training experience, one where users are supported in making informed decisions rather than being asked to plan alone or follow opaque recommendations.
Translate insights into training decisions
(too many screens to display, you’ll get the gist of it from these six.)
Health-focused apps require more information than most products to function well, but long onboarding flows often fail before users understand why the questions matter. This flow was designed to reduce early drop-off by anchoring every data request to a clear value proposition. Instead of presenting a continuous questionnaire, the flow interleaves short explanatory screens that communicate how the app uses this information to deliver meaningful insights, recommendations, and long-term progress tracking.
By tying each set of questions to an immediately understandable benefit, the onboarding process feels purposeful rather than extractive. Users are gradually introduced to the system’s core capabilities while providing the inputs needed to personalize their experience. This approach makes a moderately long onboarding flow feel lighter, builds trust early, and ensures users enter the product with a clear mental model of how their data translates into outcomes.
Miscellaneous screens
The home and workout screens are designed to orient users instantly by surfacing only the most time-critical information. The flow begins with the workout of the day, followed by body composition progress and a clear signal for when the next analysis is due, establishing rhythm and direction without exploration. From there, users move into the workout screen, where the scheduled session is reinforced through a simple weekly streak visualization, before optionally exploring additional pre-curated workouts organized by targeted muscle groups. This flow ensures users always know what to do next, while keeping secondary choices accessible but unobtrusive.
Once a workout begins, the interface shifts into an execution-first mode. The flow progresses exercise by exercise, using progressive disclosure to collapse completed and upcoming movements, keeping attention fixed on the current set. Integrated rest timers support pacing without pulling users out of the session, while an alternate exercise selection allows real-time adaptation when equipment is unavailable. This flow prioritizes continuity, enabling users to maintain training intent despite environmental constraints, without breaking focus or momentum.
The workout archive flow is designed to support continuity over reinvention. Users move from a chronological view of previous weekly plans into a focused summary of each program’s goals and structure, then choose to either set it active as-is or revise it before reactivation. By limiting actions to reuse or refine, the flow reinforces a sense of assisted control, allowing users to build on past decisions while remaining supported by the system rather than constrained by it.
moving forward
With the core flows defined, the next step is to validate the system through a focused MVP. The initial priority would be building a lightweight document ingestion and parsing layer using a constrained set of canonical fields, informed by earlier research into dependable recognition systems. This would allow the product to reliably extract body composition data and generate foundational workout and diet insights without introducing unnecessary complexity at the outset.
Once ingestion and insight generation are stable, attention would shift toward operationalizing the feedback loop. The workout editing flow becomes the primary learning surface, capturing user responses to system suggestions and using those signals to refine future recommendations. This staged approach ensures the MVP validates both technical feasibility and behavioral learning, allowing the system to progressively understand users better through repeated cycles of measurement, suggestion, and action.
