Process

Good design is the result of a clear process, not inspiration. Here's how I work.

Discovery

Every project begins with listening. I conduct stakeholder interviews, user research, and competitive analysis to build a deep understanding of the problem space. This phase establishes the strategic foundation — the 'why' before the 'what.' Atypica runs AI-powered user research before a single concept is sketched — compressing weeks of validation into hours.

Tools I use

AtypicaChatGPTClaudeGemini

Atypica runs AI-powered user research and concept validation before a single pixel is drawn. Claude, ChatGPT, and Gemini help synthesise research, map patterns, and frame the problem space.

Discovery phase — Atypica and ChatGPT illustration

Strategy

With research in hand, I define the product strategy: user personas, information architecture, and success metrics. This is where we align on what we're building, for whom, and how we'll measure its impact. No pixel is pushed until the strategy is airtight. Paraflow generates PRDs, user flows, and specs on a single canvas — keeping strategy and design permanently aligned from day one.

Tools I use

ClaudeChatGPTParaflow

Paraflow acts as an AI product design agent — generating PRDs, user flows, and specs on a single canvas, keeping strategy and design permanently aligned. Claude and ChatGPT support narrative framing and decision logic.

Strategy phase — Claude and ChatGPT illustration

Design

From low-fidelity wireframes to high-fidelity prototypes, the design phase is iterative and collaborative. I work in tight feedback loops with engineering and product teams, using design systems to ensure consistency and scalability across every touchpoint. Lovable turns designs into working React prototypes without code. Midjourney generates visual concepts and illustration assets at speed.

Tools I use

FigmaLovableParaflowMidjourneyClaude

Figma for high-fidelity design and systems. Lovable for rapid React prototyping — turning designs into working, handoff-ready products without code. Paraflow keeps UI, flows, and specs connected throughout. Midjourney for AI-generated visual concepts and illustration assets.

Design phase — Figma and Lovable illustration

Delivery

Design doesn't end at handoff. I work closely with developers through implementation, conduct design QA, and iterate based on real-world usage data. The goal is not a beautiful mockup — it's a beautiful, working product. Lovable exports production-ready frontend code directly from prototypes — closing the gap between design and engineering entirely.

Tools I use

FigmaLovableClaude

Lovable bridges design and development — exporting clean, production-ready frontend code directly from prototypes. Claude supports QA thinking, iteration logic, and documentation.