Kevin Keiper
About Resume Get in touch
Case study

NeU/X

An AI co-pilot that turns creator goals into actionable plans.

Self-directed capstone project. AI research, Figma prototype, React final build in Visual Studio with Claude Code.

Role

Solo capstone, chief build officer

Team

Solo

Timeline

August 2026

Platform
  • Figma
  • AI-assisted code
  • CoLoop + Listen Labs
  • React
  • Visual Studio
NeU/X goal-setting screen asking the creator what their goal is
Problem
Creators know their goal but not the milestones to reach it, and planning backward from a goal is repetitive work.
Opportunity
An AI copilot that works backward from a stated goal to concrete milestones, prototyped in Figma and rebuilt in React.
Result
A working React prototype whose goal-to-milestone method generalizes across creator goals, grounded in 3 research interviews.

NeU/X began years ago as a side idea to automate repetitive UX work.

A class revived it: milestones can be calculated backward from a goal, no judgment needed.

The half-joking premise: build something that makes me obsolete.

Research ran through Listen Labs, CoLoop, and Hey Marvin, mainly to learn the tools.

3 real classmates and 17 invented profiles; I judged email campaigns Priya's sharpest need.

Journey map with persona, frictions, opportunities, and AI-supported features mapped across 8 steps
Journey map: persona, frictions, and opportunities mapped across 8 steps from goal to advocacy.
Listen Labs screener and interview builder
Screener survey and interview flow, built in Listen Labs.
CoLoop AI copilot analyzing qualitative research themes
Interview responses analyzed and broken into themes in CoLoop.
Completed research responses with quality scores and summaries
3 completed interviews, each scored 5.0 on response quality.

That became two artifacts: a journey map for Priya and a record-launch process map.

Priya's journey map, mapping persona, frictions, and opportunities across the release process
Priya's Journey Map.
Record launch process map breaking down each step of a release
Record launch process.

Early screens went from Uizard into Figma: about 10 clickable screens for Priya's book launch.

The AI copilot panel evolved from a blank prompt to insight tied to the creator's last action.

Then the class ended, and it sat parked for about a month.

Low-fidelity black and white wireframe of the goal-setting screen
Early Uizard wireframe: goal-setting screen, b/w.
Low-fidelity black and white wireframe of the milestone plan grid
Uizard milestone plan, wireframed as a card grid.
Milestone dashboard with an empty AI co-pilot insights panel
Moved into Figma: initial home screen.
Milestone dashboard with a specific, contextual AI co-pilot message
From initial b/w card grid, to Figma card.
NeU/X goal-setting screen asking the creator what their goal is
Co-pilot continually responds to parent frame.
Email campaign flight check results, all checks passed
Preliminary Figma screen used to test the proof of concept.

A month later I rebuilt it in React, with about 10 milestones, email now just one.

The journey map tested the mechanic on a book launch; the rebuild, on a record deal.

The milestones change, but the goal-to-plan logic doesn't, and that's the point.

A React milestone dashboard walks creators from goal to release.

Its AI copilot reads the actual state of their plan, not generic prompts.

A campaign flight check reviews subject lines, metadata, and send timing before anything sends.

Milestone plan configuration with pre-formatted email templates
The NeU/X homescreen.
Email campaign performance overview for a milestone
Submit to distributor.
Detailed view of Priya's step two milestone
Configure email campaign.
Priya's view of the edit metadata and images screen
Priya's view of the edit metadata and images screen.
Trade-offs
The email campaign shipped as one flat step, not the fuller teaser/launch-day/thank-you sequence originally mapped out — tabled for cost, not because it didn't matter.
There's no real LLM behind the co-pilot yet, either. This is a proof of concept built to show the shape of the product cheaply, not production wiring.
The real friction was mine: not understanding React deeply enough to have an informed opinion when the build needed one from me, not just a green light.
3
Completed research interviews
8
Steps mapped, goal to advocacy

The real outcome was learning to find product-market fit, and admitting where evidence was thin.

NeU/X is a working React prototype whose goal-to-milestone method generalizes across creator goals.

That's the business value: a repeatable way to turn any goal into a concrete plan.

  • Capstone
  • AI Product
  • Figma → Code