CASE STUDY / 05

Chris Designs Things.

An AI-assisted portfolio shaped by design knowledge in Figma, then built and launched in Next.js with Codex.

AI-ASSISTED DESIGN / HUMAN DIRECTION

Chris Designs Things portfolio homepage on desktop
The finished portfolio connects product thinking, visual craft and a reusable design foundation.
ROLE
Product & visual designer
SCOPE
AI workflows, Figma, development, launch
FOCUS
Design judgement, systems, implementation

01 / FOUNDATION

Start with a foundation. Make it unmistakably yours.

I wanted a portfolio that could explain the range of my work without making it feel disconnected. AI helped explore the content, automate repetitive setup and compile reusable building blocks into an initial foundation. I then styled and refined those elements directly in Figma, using typography, hierarchy and composition to give the site a considered identity.

DESIGN PRINCIPLE

The brief comes first.

Define the audience, purpose and constraints before asking AI to generate a layout. A useful foundation gives the designer something concrete to evaluate and improve.

02 / CONNECTED WORKFLOW

Three ways of working. One continuous process.

I moved between Chat, Work and Codex as the project needed conversation, action or implementation. Each had a practical role in my workflow, with Figma carrying the visual decisions between stages.

01 / CHAT

Think and frame

A conversational space for exploring direction, refining copy and making a brief clearer. I used it to work through the story, page structure and questions before committing to a design.

02 / WORK

Act on the brief

A task-focused way to use files and connected tools to produce something reviewable. I used it with Figma to assemble components, apply tokens and speed up repetitive layout work, then styled and refined the result myself.

03 / CODEX

Build and iterate

A coding agent that can inspect and change project files. I used it to build the Next.js website from my Figma designs, resolve implementation issues and refine behaviour through technically informed requests.

03 / MCP + CONTEXT

Connect AI to the work.

Model Context Protocol, or MCP, is a standard way for an AI assistant to access connected tools and context. In this project, Figma connections gave the agent access to frames, components and variables. The design became a reference it could inspect and act on, with my instructions defining what to preserve and what to change.

FIGMA

Design reference

Frames, tokens and components

MCP

Tool connection

Connected design context

CODEX

Next.js website

Implementation and refinement

The connection carried context. My review determined whether the result respected the design.

INSTRUCTIONS

Prompt

The brief and constraints you give the assistant.

TOOL CONNECTION

MCP

A standard connection between the assistant and tools.

DESIGN VALUES

Tokens

Named values for colour, typography and spacing.

BUILDING BLOCKS

Components

Reusable building blocks with consistent properties.

04 / DESIGN DIRECTION

Design knowledge shaped the foundation.

My education in design principles helped me bring structure and character to the initial foundation. I styled and refined the AI-assembled building blocks directly in Figma, using typography, hierarchy, proportion, contrast and spacing to create a cohesive visual direction. Shared tokens and reusable components kept those decisions consistent across the site, while desktop and mobile layouts adapted the same story to different screen sizes and reading experiences.

MOBILE / READING ORDER

Chris Designs Things homepage with the mobile content reading order
Mobile: content stacks in a deliberate reading order, with controls and visuals sized for the smaller canvas.

DESKTOP / SPACE + HIERARCHY

Chris Designs Things homepage with its desktop typography and composition
Desktop: larger type, generous space and wider compositions give the work room to speak.

05 / DEVELOPMENT + MOTION

Technical fluency made the prompts actionable.

I developed and launched the site in Next.js using Codex. My HTML/CSS background helped me give precise instructions about layout, component structure and responsive behaviour, then spot and correct implementation issues. That knowledge made the AI tools more effective and efficient, substantially reducing overall development time. I continued refining the build and adding animations through focused prompting, adjustment and review.

01 / STRUCTURE

Name the layout

Describe containers, alignment and sizing so the build follows the design’s underlying rules.

02 / RESPONSIVE

Specify behaviour

Explain what should stack, wrap or change as the available space becomes narrower.

03 / MOTION

Tune the details

Adjust transitions, reveals and hover behaviour to support hierarchy and the site’s visual direction.

04 / REVIEW

Keep refining

Compare the implementation with the design and correct the specific places where it drifts.

06 / LAUNCH + REFLECTION

A launched website. A repeatable way to work.

The finished portfolio brings my product and visual work together through shared foundations, responsive layouts and purposeful motion. The process gave me a repeatable way to move from a brief to a launched website, with AI accelerating production and my design experience guiding the decisions that shaped the result.

01 / RESULT

Shared foundations

Tokens and components guide the design.

02 / RESULT

Responsive layouts

The story adapts to desktop and mobile.

03 / RESULT

Purposeful motion

Animation supports the visual direction.

04 / RESULT

Next.js website

Developed and launched with Codex, with design-led refinement.

The value of this workflow came from knowing what to ask for, what to keep and what to refine. AI helped me move faster, while an understanding of design principles and web development kept the final product purposeful and distinctly my own.

CHRIS TRAVAGLINI / CASE STUDY 05