Approach
How we work.
Designers make the decisions. AI does the building. Real users confirm whether we got it right.
Process
Four stages, from idea to a product people use.
Discover
We learn who your users are and what they need, and agree on the problem worth solving.
What happens
Interviews and field research with real users, a review of your current product, user personas, and a solution architecture that shows how the pieces fit together.
What you receive
- Research findings
- User personas and needs
- Solution architecture
- An agreed problem statement and success measures
Prototype
We design the experience, and AI turns it into a working prototype within days.
What happens
User flows and screens on your design system, built by AI into an interactive prototype that behaves like the real product.
What you receive
- Experience design in Figma
- A working, interactive prototype
Validate
We test the prototype with real users before you commit to a full build.
What happens
Usability sessions with the people who will use the product. We fix what doesn't work, then test again.
What you receive
- Usability findings
- A refined, validated prototype
- A clear decision on what to build
Build and scale
We build the first release with AI, under your rules, and set up the team that sustains it.
What happens
AI-assisted engineering, governed by specifications, skills and independent checks. Where needed, we also set up your design team and support hiring and training.
What you receive
- A first release (MVP)
- Figma files and code your engineers can extend
- Your Design Constitution
- A team ready to take it forward
Roles
Clear roles, for people and for AI.
Designers lead
Research, personas, experience design, architecture and the rules: every decision that requires judgment.
AI builds
Prototypes, code and variations, following your Design Constitution and skills.
A checker verifies
An independent check on every task, against agreed quality gates. Nothing is complete until it has been verified.
AI-native workflow
From your design system to working code.
We connect Claude to Figma through the Model Context Protocol (MCP). Working directly with your design system, it produces Figma files, interactive prototypes and React code your engineers can extend.
- Figma files on your design system
- Interactive prototypes
- React code
Design Constitution
One set of rules for designers and AI.
A Design Constitution captures how your product should look, behave and read: your design system, interaction patterns, brand rules and accessibility requirements. Designers and AI tools work from the same file, so output is consistent from the first draft.
# Design Constitution: Your product
## Components
- Use design system components. Never build a new button.
- One primary button per section.
## Accessibility
- Text meets WCAG 2.2 AA contrast.
- Every form field has a visible label.
## Voice
- Short sentences. Plain words.
- Buttons say what happens: "Send enquiry", never "Submit".Research methods
How we learn what users need.
- Interviews
- One-to-one conversations about goals, habits and frustrations.
- Field studies
- Observing people as they work, in their own environment.
- Usability testing
- Real people completing real tasks on a prototype or live product.
- Card sorting
- How people group and label information, to shape navigation.
- Journey mapping
- Every step, channel and handoff, end to end.
- Personas
- Who your users are, grounded in research rather than assumptions.
- Analytics review
- What usage data shows, and where it differs from what people tell us.
See how this would work for your product.
We reply within 2 working days.