unakin

Study U-01 — Sedulous

A readiness platform for tactical athletes.

Case study · Client work

Client
South Florida Tactical Athletes
Commissioned by
Dr. Arnold T. Stocker
Year
2026
Role
Design + build
Status
In beta

Sedulous is a mentor-guided readiness system. A mentor writes the sessions; the athlete follows them, on a phone or a watch. It is built on one idea: the gap between what an athlete claims and what gets verified is the measure. And it is secure from the inside out: its servers never learn an athlete’s name.

Plate 01 · The teaser · 10 seconds

The man who asked for it.

The client

Sedulous was commissioned by Dr. Arnold T. Stocker, founder of South Florida Tactical Athletes. He is a Pararescueman, an 18D Special Forces combat medic and a Lt. Colonel, with four decades of service.

Sedulous puts a mentor’s method into a product: write the work, watch it done, and count a result when someone has witnessed it. It is built so that method can reach more athletes than one mentor can stand beside.

An athlete today. An operator tomorrow.

The secret sauce · Security from the inside out

Some of the people training in Sedulous today will go on to serve: as military personnel, as Special Operations candidates, as operators. For them, a training record with a name on it is a liability years from now, long after anyone remembers it was kept.

So Sedulous is secure from the inside out. It is built never to hold a name. This is not a privacy setting someone can switch off, or a wall around data that should not be there. It is a decision about where the data lives.

01 Call signs
Athletes appear by call sign everywhere in the product. Not by name.
02 The vault
Real names live in a vault on the mentor’s own device. They never reach the server.
03 The database
A copy of the database holds call signs and performance, and nobody’s identity.
04 The route
A run’s GPS route is never saved or sent. Sign-in is by passkey, so there is no password to leak.

Why it holds

  1. Not holding names is the strongest protection there is. You can’t leak, sell, or be forced to hand over what you never had. Most products do the opposite and then build walls around it.
  2. Not saving GPS routes is the right lesson learned. In 2018, a fitness app’s public heat map exposed the locations of military bases through soldiers’ runs. Leaving routes out avoids that whole class of problem.
  3. Passkeys remove passwords and most phishing. There is nothing to guess, reuse or be tricked into typing.
  4. The AI mentor never sees a name. It sees call signs, performance and what the athlete declares. From those it builds around how each athlete is actually performing, so the work stays scaled and targeted to their goals, and changes as their performance does.

Prove it cheaply, before it has an app.

01 Draft · What we thought

The hard question was never the screens. It was whether a mentor and an athlete would keep a loop going: write the session, do the session, compare what was claimed with what was verified. Native apps are the expensive way to find that out.

So the first version was a web app: one codebase, no store review, and a change is live the day it is made. Passkeys work in the browser, so nothing about the product needed a native app to prove the loop. The web app carried Sedulous through every round of testing and rethinking. Only then did the money go into iPhone and Android, and by then we knew exactly what to build.

Plate 02 · On the phone · captured from the public demo

Give it a spine.

02 Form · What everything hangs from

One web app is the product. The phone apps are a thin shell around it, so reaching the App Store and Google Play was a matter of packaging, not rewriting screens. Change the product once and every phone has it.

The same structure decides what the server is allowed to know. That turned out to matter more than any feature, because of who the athletes are and who some of them will become.

No.Made ofWhat it does here
M-01AstroThe web app, rendered on the server
M-02Web components + plain CSSThe interface, with no framework to keep up with
M-03Supabase (Postgres)Data, with row-level security on every table
M-04PasskeysSign-in. No passwords
M-05CapacitorThe iPhone and Android shell
M-06SwiftUI · Wear OS · Connect IQThree native watch apps
M-07ClaudeThe AI mentor that adjusts the day
M-08Shipmate · UNAKIN StudioOur own tools: deploys, and the invitation emails

Nothing to think about. Just the work.

How it works

01 Connected
Everything is interconnected. The phone, the watch and the web carry the same session.
02 Guided
Nothing to think about. The mentor plans it and the watch walks you through it.
03 Watched
AI scans everything. Humans review everything.
04 Yours
It scales to your actual performance, so that you meet your actual goals.

Plate 03 · On the wrist · Apple Watch and Garmin

Ship it, everywhere they train.

03 Finish · What shipped

In the two weeks from 5 November, Sedulous was rebuilt around passkeys and taken from one web app to six surfaces: Garmin, Apple Watch, Android watches, iPhone, Android and the web. The watches are native, because a session on the wrist has to work mid-run, with a guided step-by-step view and a pace coach.

It could move that fast because the thinking was already done. On the web a change costs minutes, not a store review, so every idea had been tried, kept or thrown out there first. By the time money went into iPhone, Android and the watches, the screens and the rules were settled, and the apps only had to carry them.

No.SurfaceHow it is builtWhere it is
S-01Web appWhere every idea was tested first. Installable to the home screenLive
S-02iPhoneA thin shell around the live web appTestFlight
S-03AndroidThe same shellPlay testing
S-04Apple WatchNative, SwiftUITestFlight
S-05Android watchesNative, Wear OSPlay testing
S-06GarminNative, Connect IQStore beta

The log · two weeks

  1. Sign-in is rebuilt around passkeys. No passwords. The revamp starts here.
  2. iPhone build in TestFlight, Android build in Play testing.
  3. Watches pair with the app.
  4. An end-to-end test loop covers the whole product.
  5. Garmin app published to the Connect IQ Store as a beta.
  6. An AI mentor that adjusts the day, and releases that ship with one command.

By the numbers · 5 to 17 November 2026

281
commits to the product in thirteen days
207
more to the phone and watch apps
6
surfaces, one product
16
steps in the end-to-end test loop

Tell us what you're building.

Start a project