Skip to content

Work · Digital / Building

AI-Assisted Web Development

Turning creative and professional requirements into working digital interfaces with AI assistance, under human control.

Screenshot of Fomzo Paris — a client-facing beauty-services website built with AI-assisted web tools.

01

Context

A portfolio strand covering how AI-assisted development is used to take creative and professional requirements through to working, maintained web interfaces.

02

Question / Challenge

How much of an interface can be built with AI assistance while structure, content, testing and visual direction stay under human control?

03

My role

Requirements, information architecture, content, visual direction, review and testing.

04

Process

  1. 01Define the audience, structure and content before any code is written.
  2. 02Set the design system and typographic rules, then hold implementation to them.
  3. 03Use AI assistance for implementation of defined components and layouts.
  4. 04Read, test and correct the output; rewrite anything that does not match the intent.
  5. 05Check accessibility, responsiveness and content accuracy before release — keyboard focus, skip link, semantic structure, reduced motion, and layouts at 1440, 1024 and 390px.

05

AI contribution

Implementation assistance across markup, styling and component structure.

06

Human judgement

Architecture, copy, design system and every accessibility and accuracy decision. Generated copy containing unverifiable claims was removed.

Contribution map

Human

  • Requirements and information architecture
  • Content and copy
  • Art direction
  • Accessibility review
  • Testing and correction
  • Final decision

AI

  • Implementation assistance
  • Component drafting
  • Structuring

07

Output

Working sites, including this portfolio — built in Lovable with AI-assisted implementation on a TanStack Start and TypeScript architecture, with responsive layouts checked at 1440, 1024 and 390px, a skip link, semantic structure, visible keyboard focus and reduced-motion support — alongside an earlier personal portfolio and Fomzo Paris, a client-facing beauty-services website.

08

AI disclosure

AI systems used
Lovable, with AI-assisted implementation across a TanStack Start and TypeScript codebase.
Why AI was used
To move from defined requirements to working interfaces without a development team.
What was verified
Rendered output, responsive behaviour at 1440, 1024 and 390px, keyboard access and focus states, reduced-motion behaviour, and the factual accuracy of all site copy.
What AI got wrong
Invented copy and unsupported claims, inconsistent use of the design system, and accessibility details such as labelling and focus states.
Rejected or changed
All fabricated content; any styling that broke the established design tokens.
Human contribution
Structure, content, visual direction, accessibility and final review.
Ethics & authorship
AI assistance in building these sites is stated openly. No client outcomes or metrics are claimed.

Tools

  • Lovable
  • TypeScript
  • TanStack Start
  • Tailwind CSS

Skills demonstrated

  • Information architecture
  • Design systems
  • Accessibility
  • AI-assisted implementation
  • Testing and review

Gallery

Screenshot of Violetta Liszka's earlier personal portfolio website.
Earlier personal portfolio — AI-assisted web build.
Screenshot of the Fomzo Paris beauty-services website homepage.
Fomzo Paris — client-facing beauty-services website built with AI-assisted web tools.

Links

Reflection

AI assistance shortens implementation; it does not shorten the decisions. The review step is where the work actually is.

Related projects