Please ask Christopher for the password, or get in touch.
Case Study · 2025–2026
Fenix Forge: building an AI-powered coaching platform
UX Cerebro is its confidence reclamation engine — designed for UX designers rediscovering what they already know
Role
Solo designer, product owner, and builder
Timeline
February 2025 — ongoing
Stack
React + Supabase + Netlify + Anthropic API
Status
Live in production · 18 clients · 3× faster onboarding
Phase 1 Discovery — a live coaching exchange, one question at a timeThe confirmation gate — the student decides when Phase 1 is done
Group Sessions — recorded live coaching calls, archived inside the platform for students to revisit
18
coaching clients
3×
faster onboarding
$2,400+/yr
vendor cost eliminated
The Problem
Why I built this for myself first
I have coached over 158 UX designers and researchers since 2020. For years the delivery mechanism for my methodology was friction-heavy: record a video, build a Google Slides deck with prompts, direct clients to their own AI tool, have them paste output into a document, debrief on a call.
It worked. But I was rebuilding the wheel for every student, context-switching across too many tools, and absorbing the overhead of a platform that had no live-session integration and created content lock-in risk.
The deeper problem was invisible to most people but obvious to me: my students were arriving at sessions underprepared. They were bringing raw confusion rather than processed clarity. I was spending the first 20-30 minutes of every session doing work the student should have done before we got on a call.
I was the instrument of their reflection instead of the amplifier of it.
I needed to change the infrastructure.
What I Already Knew
A pattern I'd already seen 18 times
I had been running clients through a 4-phase self-discovery methodology for over a year before building any of this. The framework was proven empirically. I had watched it produce results across 18 direct coaching clients and observed the same pattern repeat across all of them.
This told me the methodology was sound. The delivery mechanism was the gap.
The pattern that proved the methodology
The pattern: when a student goes through the excavation exercise, they experience what I call career amnesia reversal. They rediscover skills, projects, and proof points they had normalized into invisibility. One client, Shoji, had 10 years of freelance creative experience he had dismissed as irrelevant. After Phase 1, he messaged me with a series of epiphany GIFs. He had not acquired anything new. He had stopped hiding what he already had.
The Design Opportunity
Naming what students can't see themselves
The insight that changed direction came from watching my own clients closely. The moments of highest value were not when I gave them information. They were the moments when I named something they could not see from inside their own experience. The AI coaching conversation could replicate that, if the system prompt was built correctly.
“How do I create the conditions for self-discovery at scale?”
Guiding Principles
Three beliefs, every decision
Every decision in this build was made against three principles.
Excavation over invention
The product surfaces what already exists in the student. It does not generate a new version of them. Every coaching question goes one level deeper. Every output is built from the student's own language.
Pacing is the product
Reflection checkpoints are not loading states. They are the work. Phase transitions are not forced. The slowdown at the right moment is the value, not a bug.
The artifact belongs to the student
Confidence borrowed from a coach disappears when coaching ends. Confidence rediscovered from one's own history travels permanently. Every decision ensures the student leaves with something they made.
What I Built
Two systems, one constraint
Two connected systems: UX Cerebro, the 4-phase coaching engine, and Fenix Forge, the platform that houses it.
Phase 1: Discovery
A live AI coaching conversation. One question at a time. The system prompt is trained on my methodology: warm but probing, pushing deeper without redirecting. After 6 meaningful exchanges, the student sees a coaching nudge: “Do you feel there's more to explore, or are you ready to move to Phase 2?” Two choices. No forced progression.
Phase 2: The 7 Power Moves
A single bounded synthesis call. Claude reads the full Phase 1 transcript and generates 7 named strategic assets: the student's unfair advantages, articulated in their own language. Each Power Move includes the insight, why it matters, the strategic edge, and when to deploy it. This phase has produced the most emotionally significant moments in testing.
Phase 3: Activation
Transforms the Power Moves into a tactical playbook: interview story arsenal, portfolio narrative strategy, 30/90-day career roadmap, and power networking frameworks. Specific to the student's background and goals, not a generic template.
[Screenshot: Phase 3 Activation playbook]
Phase 4: The Brand Playbook
A comprehensive living document exported as a Word file the student owns, edits, and updates quarterly. Sections include: executive summary, origin story, all 7 Power Moves with deployment language, activation strategies, LinkedIn and email templates, a quick reference guide, and a personal UX manifesto.
[Screenshot: Phase 4 Brand Playbook document]
Fenix Forge: the coaching platform
The infrastructure layer that houses UX Cerebro and the broader coaching program. Built on React + Vite + Supabase + Netlify, it replaces a third-party vendor costing over $200/month — $2,400+ a year — that created content lock-in risk. The platform is designed around one constraint: students arrive at live sessions 80% prepared. The platform handles excavation, synthesis, and documentation. I handle belief transfer.
Key Decisions
The reasoning behind five choices
Built vs. bought
Why I built instead of bought
Every third-party platform I evaluated optimized for content delivery. None were built for the interaction pattern my methodology requires: an AI coaching conversation that excavates rather than instructs. I could not buy what I needed. I had to build it.
Claude as backbone
Why Claude as the AI backbone
The coaching persona required a model that could hold nuance, probe without prescribing, and stay in a coaching register across a long conversation. The system prompt I wrote is the product. The model is the instrument.
Two-button nudge
Why the Phase 2 nudge has two buttons
The original spec called for a button appearing after 6 messages prompting the student forward. I changed it because forced progression contradicts the core coaching philosophy. The student decides when Phase 1 is complete.
Word doc vs. PDF
Why Word doc over PDF
The Brand Playbook is a living document. Students are meant to return to it quarterly, update it with new wins, and reference it before every important career moment. A PDF is a snapshot. A Word doc is a tool.
Function first
Why function before aesthetics
The first live version was deliberately rough. I needed to know the coaching conversation worked before investing in visual polish. I ran myself through the full flow before showing it to a single student.
Outcomes
Where things stand
18
coaching clients
3×
faster onboarding
60–70%
session efficiency gain
$2,400+/yr
vendor cost eliminated
Students now arrive at sessions with processed clarity, reducing session hand-holding by an estimated 60–70%. Onboarding — Phase 1 through Phase 2 — used to take about three weeks of back-and-forth over email and slide decks. A dedicated student now gets through it in under a week: roughly 3× faster. The platform also replaces $200+/month ($2,400+ a year) in third-party vendor cost with self-owned, extensible infrastructure — live with Phase 1 and Phase 2 functional.
What I Learned
Methodology and medium aren't the same thing
Insight
Methodology vs. medium. I had a proven coaching methodology and 18 clients who could confirm it worked. What I did not have was a medium that respected the pacing the methodology required. Every platform I tried optimized for throughput. My methodology optimizes for depth. Those two things are in direct conflict.
Implication
Building my own platform was not vanity. It was the only way to protect the integrity of the method.
Insight
Running myself through it. Running myself through Phase 1 the night before a product interview, and surfacing a genuine insight about my own career narrative, was one of the most clarifying professional experiences I have had.
Implication
The product worked. Not because I built it correctly. Because the methodology it contains is true.
What's Next
The rest of the build
Phase 3 and Phase 4 build: completing the full 4-phase loop
Supabase auth: two-tier access so students see their own sessions and I see everything as coach admin
Phase 5: LinkedIn outreach protocol generating personalized messages from the student's Brand Playbook
Document upload: returning students can upload previous Cerebro output and reconstruct context for a new session
