Brendan FlynnProduct Operations · Service Design · Customer ExperienceDownload CV
← All selected work

INDEPENDENT PRODUCT

The Loop

A privacy-first athlete product combining hands-on delivery, database-enforced boundaries and deliberately narrow AI support.

Product implementedValidation pending
Role
Independent product lead and builder across discovery, interaction design, architecture, data boundaries and implementation.
Maturity boundary
The product architecture, core loops, privacy controls and automated checks are implemented. There is no public user-impact claim: the first external cohort and behavioural validation remain future gates.
ATHLETEPrepare · Capture · ReflectPrivate experience
EVENT-ROOTED COREIdentity · Evidence · LifecyclePostgreSQL · RLS · guarded functions
COACHSchedule · Participation · PatternsAggregate signals only
Athlete free text never crosses the coach boundary
Two roles share an event lifecycle, not private reflection data.

CONTEXT

Athletes need a useful rhythm around preparation and reflection, while coaches need team-level signals without access to the private thoughts that make honest reflection possible.

MY CONTRIBUTION

I shaped and implemented an event-rooted mobile product with separate athlete and coach surfaces, database-enforced privacy, deterministic safety logic and narrowly scoped AI support.

WHAT IS TRUE NOW

A substantial pre-release product exists across athlete, coach and platform surfaces. Engineering quality is evidenced; external athlete behaviour and performance impact are not yet validated.

  • Athlete ownership
  • Database-enforced privacy
  • AI with boundaries

CONTEXT & PROBLEM

Reflection only works if privacy and usefulness coexist.

Athletes need space to prepare deliberately, capture an honest first reaction and carry one useful lesson forward. Coaches need enough team-level context to support participation and patterns, but not access to the athlete's private record.

That made trust part of the product architecture rather than a policy layer to add after the experience was designed.

DISCOVERY & CORRECTION

Model the sporting event before modelling the screens.

The central architectural correction was to make each game or training session the durable root of the system. Athlete and coach experiences became different views of the same lifecycle, while derived patterns retained a path back to their evidence.

SOLUTION SYSTEM

Two product surfaces; one carefully bounded lifecycle.

React Native provides the mobile product. Expo accelerated prototyping; an EAS-built TestFlight candidate and physical-device QA form the next release gate, while the longer-term mobile platform remains an explicit decision rather than an assumption. Supabase, PostgreSQL and row-level security enforce identity, provenance and access boundaries. Named attendance remains distinct from private reflection, and coach insight is aggregate-only.

  • Athlete: prepare, capture, reflect and carry learning forward.
  • Coach: schedule, see participation and receive thresholded team patterns.
  • Platform: preserve identity, evidence and permissions at the data layer.

AI & SAFETY

Use AI where variation helps, not where certainty matters.

Deterministic logic owns identity, permissions, evidence thresholds, crisis routing and coach access. AI is bounded to concise acknowledgement and guided reframing, with server-held credentials and a static fallback.

Personal patterns are based on the athlete's own history and prefer silence to a weak claim. The athlete can reject a pattern that does not feel true.

DELIVERY EVIDENCE

Engineering proof is necessary; it is not product validation.

The implemented system has automated checks, continuous integration and explicit database controls. Those are meaningful signs of technical delivery discipline, but they do not prove that athletes will return, reflect honestly or improve performance.

The next gate is observed use with a bounded cohort and pre-defined decision criteria. The product should earn further investment through behaviour, not build volume.

SUPPORTING PROOF

  • The private implementation repository was reviewed for product surfaces, architecture, tests and continuous integration.
  • Public evidence is restricted to sanitised system descriptions; no private code, customer data or safeguarding detail is exposed.
  • Maturity is stated as pre-release internal testing, not market validation.

WHAT I CARRY FORWARD

The strongest product decision was recognising validation debt.

Implementation moved ahead of athlete observation. Correcting the data model and privacy boundary improved the system; the remaining work is deliberately behavioural—ship narrowly, observe honestly and let athletes change the theory.

Explore all selected work