brewing…
Fluxi · concept → MVP
Talented designers lose roles to unclear stories, not weak work. Fluxi is an AI-first portfolio builder that turns scattered project notes into a structured, job-ready case study — coaching the narrative at each step, not just the layout.
Overview
Fluxi is an AI-first portfolio builder for designers. Existing tools optimize for layout flexibility and hand you a blank canvas exactly when you need a narrative framework. Fluxi turns your raw inputs — notes, artifacts, decisions — into a structured draft, then coaches you through refining it into a clear, credible case study.
Your raw inputs
A structured case study
I took Fluxi from a problem and real user pain to a working MVP prototype, using an AI-first process to move fast — testing the storytelling experience itself, not just designing screens for it.
The problem
Talented designers do the work — then struggle to articulate it. The visuals are strong, but turning the messy reality of a project into a clear, structured case study is a different skill entirely, and most were never taught it. Great work, unclear stories. Three things make it hard:
From my own research — 20+ designers across interviews and a survey, plus conversations with hiring managers on what gets someone shortlisted.
The question
How might we help designers turn strong work into a story that lands in the seconds a reviewer actually spends?
Who it's for
Design student · early-career
“I don’t know what recruiters want to see — so I freeze.”
Goals
Frustrations
Mid-level designer · 2–5 years
“My work is strong — my case studies just read like feature walkthroughs.”
Goals
Frustrations
Research & validation
I grounded Fluxi in real input, not assumptions — 20+ designers, through interviews and a survey, on where portfolio storytelling breaks down, plus hiring managers on what actually gets someone shortlisted. One pattern held across both: clear structure leads to a clear story. The problem wasn’t talent or effort; it was the absence of a narrative scaffold at the moment people needed it most.
The solution
Fluxi replaces the blank canvas with a guided flow: you bring the raw material, and it builds the scaffold with you.
Context in
Notes, files, even voice memos.
Fluxi thinks
Clusters and finds the meaning.
A base draft
A structured, story-ready start.
You refine, guided
Sharpen it with AI coaching.
Guidance dials from full hand-holding to fully hands-off — so the same product fits a beginner like Mia and a time-pressed pro like Aarav.
Why not just ask ChatGPT or Claude?
Fair question — and general models are genuinely capable now, with project context and memory. But they’re general. To get a strong case study out of one, you still have to be the expert: you have to know what makes a case study land, how recruiters read, and exactly how to ask. That expertise gap is what freezes designers. A capable blank box is still blank.
Fluxi puts that expertise inside the product. It knows what a design case study needs. It walks you through it, so you never engineer a prompt. And it coaches in a way that sharpens your voice instead of flattening it. The model’s intelligence is table stakes — the storytelling is the product.
A model can write anything. It can’t tell you what your story should be — that’s the part Fluxi is built for.
Key design decisions
Show only what’s needed at each step.
Instead of a full editor up front, which is exactly what freezes people.
Pre-built narrative components to build from.
Instead of a blank canvas that offers freedom but no framework.
Real-time feedback that pushes for clarity and credibility.
Instead of generic AI text that sounds fake or off-voice.
Prompts with intentional boundaries that prevent over-explaining.
Instead of open-ended fields that invite rambling, not stories.
Prioritization
I mapped every feature my research surfaced against impact and effort. The MVP is only what sits in the top-left corner — high impact, low effort — and everything else earned a later lane, on purpose.
High impact · Low effort
High impact · High effort
Low impact · Low effort
Low impact · High effort
The hardest cut
The sharpest finding in my research wasn’t “I can’t write it.” Designers write specs and research readouts constantly. It was quieter and more specific: they couldn’t tell whether what they’d written was any good. Not the prose — the substance. Was the impact real or asserted? Was the role clear? Was the outcome sourced, or had it drifted into something that sounded right?
That’s a gap general AI is uniquely bad at closing. Ask a model to review your case study and it will tell you it’s strong. It’s agreeable by default, and agreeableness is the opposite of what this moment needs.
So the feature I most wanted to build was a readiness Scorecard: every section of a draft labeled against what you actually supplied — sourced, thin, or missing — with follow-up questions generated only where the gaps are. Not a grade. A coverage report. The product’s job would be to tell you honestly what you haven’t earned yet.
I cut it from the MVP anyway.
It needed traceability from every claim in the draft back to the input that produced it — real provenance, not a label — and getting that wrong is worse than not shipping it. A tool that tells you a section is thin when you did supply the evidence loses trust permanently, in one interaction. High impact, high effort, and the effort is the kind you can’t fake.
So the MVP proves the smaller claim first: that a guided flow gets designers from raw material to a structured draft at all. The Scorecard proves the larger one — that a portfolio tool can be honest with you — and it’s what v2 is being built around.
Every competitor helps you format a portfolio. The bet is that the one people actually need helps them find the story they already have — and tells them plainly what’s missing.
Challenges
The AI-first process
I didn’t just design an AI product; I built it with an AI-first process, matching the tool to the phase — optimizing for speed early and quality later.
MVP phase
Fast UI exploration — iterate on flows without engineering.
MVP phase
Narrative testing — pressure-test the storytelling logic itself.
Future phases
Production refinement — move toward a buildable implementation.
Future phases
Complex reasoning — the harder narrative and product logic.
What surprised me
Building beat mocking
My first time vibe-coding a product — the surprise wasn’t speed, it was leverage. I could put working behavior in front of people in a day and iterate on how it felt, not on a picture of it.
AI is a design material
The real design work moved from screens to the conversation — what to ask, when to step in, how much to show. The prompt became the canvas, and getting it right mattered more than any single layout.
The real question AI forces
When a model can build almost anything in one run, “can I build it?” stops mattering. The question becomes what the product does that the model alone can’t — a product, or just a wrapper on an agent? That question drove every cut.
See it in action
What's next
The MVP — guided intake, importing, and the narrative engine — is in testing with real designers now. v2 is already in progress, with more of what makes a portfolio land: a readiness Scorecard, role-based portfolio sets, and richer guidance.
Want to see where it’s headed? Get in touch.
Reflection
Fluxi is still being built
If you’d like to hear where Fluxi’s headed or talk AI-first product design, I’d love to connect.