Implementing the interface myself kept me designing within real constraints: I could feel the cost of a pattern before committing to it, and nothing was lost in handoff because there wasn't one.
TESTING THE CORE HYPOTHESIS
Can a review on the frame replace one buried in chat?
01Scope
The smallest product that could test it.
Frame-level review
Shared mistake library
One coach ↔ one athlete
02Learn
What the alpha must show — 10 coaches, real athletes.
Coaches complete a review without switching to chat
Athletes act on a fix without asking what it meant
Coaches reuse saved mistakes instead of retyping them
Open: where the workflow breaks, and what needs automating
03Decide
Decisions waiting on those answers.
Coach discovery & matching
Subscriptions & payments
Progress across reviews
Scale & collaboration
The product model
Everything leads to one shared review.
Each role prepares on its own — then the two meet at a single moment.
Coach
Athlete
Shared start
Create a profile
01Add your certificates to your account
02Get request from athlete
03Pick mistakes and create a review
01Upload clip
02Pick coach
03Send request to your coach
Before
each role works alone
After
the outcome for each
The one point they meet
The review
The coach’s review lands on the athlete’s frame.
Coach · outcome
Get paid for the review
04Get review
Athlete · outcome
Get better and progress
Early exploration
The workflow came first. The interface followed.
Sketched end to end before a single pixel — the product is already recognizable here.
The end-to-end experience
From sign-up to review, in one loop.
The athlete sets it up, the coach resolves it on the clip, and every fix comes back in a single feed.
Complete your profile
Add your details and training goals.
Pick a coach
Filter, then choose a specialist.
Send the clip
Upload a training clip.
The coaching payoff
Every reviewed session is ready to explore.
Open any session to explore the detailed fixes, coaching notes and drills the coach left behind.
Review feed
Reviewed sessions, ready to open.
Detailed review
Frozen frame, marked joints, and the fix.
The end-to-end experience · the coach
The same loop, from the coach's side.
The coach gets verified, accepts an athlete, and resolves the clip — mistakes pulled from the shared mistake library, the review sent in one place.
Get verified
Upload qualifications for review.
Accept an athlete
Read the request, then confirm.
Open the session
Pick a submitted clip to observe.
The coaching payoff
One clip in, structured review out.
The coach marks the movement, pulls named mistakes from the shared mistake library, and sends it — the athlete receives a structured how-to-fix, not a loose note.
Annotate & flag
Mark joints, pick mistakes, send.
Delivered to the athlete
A structured, reusable how-to-fix.
Product decision · grounded in research
Why reviews start with mistakes — not comments.
Research insight
Coaches identify the movement problem before explaining it.
→
Product decision
The review starts from the shared mistake library.
1Select from shared library
Personalise feedback2
For coaches
·Less repetitive writing
·Consistent reviews
·More time for athlete-specific guidance
For the product
·Shared terminology
·Labelled examples
·Growing knowledge base
Every named mistake makes the next review faster to explain.
The system behind it
One system — from identity to implementation.
I designed a reusable product language rather than isolated screens — identity, components and interaction patterns as a single vocabulary.
01Identity
Designed inside the product — it set the language the interface speaks.
Aa
Fixed
Action
Mistake
02Components
Built once, reused across every feature — a few parts carry the whole product.
Verified coach
Knee valgus on drive
Learn more
0:28
03Patterns
The product reuses interaction logic, not just visual elements — the same behaviour repeats across features.
Search mistake→Results open→Selected
Pick tool→Active state→Marker placed
Upload clip→Processing→Ready for review
Credentials→Verified badge→Shown in search
04One language across every touchpoint
The same system carries the coach workspace, the athlete experience, discovery and verification.
Three problems changed the product. None came from a brief.
We scoped a deliberately small MVP, reviewed the workflow internally with coach feedback, and let recurring problems reshape it.
Hypothesis·Prototype·Internal review·Reusable system
01
Observed
Freehand marks read differently every time.
Marks became a shared vocabulary
Reviews can stay reusable and consistent.
02
Observed
Play-and-pause overshot the frame that mattered.
Reviewing became frame-accurate
Coaches can land on the exact moment.
03
Observed
Reviewing clips was only part of a coach’s job.
A review tool became a coach workflow
Coaches can run their whole practice in one place.
The first idea wasn’t the finished product. Three problems reshaped it.
What exists today — and what's still open.
The responsive web app, its component library and the landing page exist and work. Final internal testing is in progress; alpha with coaches and athletes is the next step, and the decisions below it stay open until then.
Design isn't the interface. It's the decisions that survive implementation.
Next: alpha testing with coaches and athletes will evaluate the review workflow and inform the annotation toolkit, collaboration model and product scope.
