Evidence, not taste.
Every major decision traces back to something a real user said or did. When someone asks why a screen is like this, the answer is never that we liked it better.
Process
How we work
Every project runs on four steps: understand, design, test and build. We research with real users, design on a reusable system, test before launch, and build or hand off to your engineers. The steps stay fixed; the methods inside them change with the product.
Step 01
Understand
We start by listening: to the people behind the product, to its users, and to the data. For AI products we also review the model's real output, not the demo.
- The vision: what the product should solve and what it should make people feel.
- The business model: how the product creates value, because design that ignores it does not survive.
- Research with users: interviews, live focus groups of up to 60 to 80 participants on larger projects, and surveys to check the signals at scale.
- An audit of the current product and teardowns of competitor flows.
- Synthesis: we transcribe, split what people said into small insights, group them into an affinity map, and distil the few findings that drive design.
What comes out
- Key insights with the evidence behind them
- Personas grounded in real users
- A priority matrix of what to solve first
Step 02
Design
We move from insights to flows, wireframes and interface, building a reusable design system as we go.
- Information architecture, where the product needs it.
- User flows, drafted and pressure-tested before any screen.
- Wireframes, then a mid-fidelity direction, reached fast so more time goes to polish and testing.
- High-fidelity interface on an atomic design system: tokens, components and patterns.
- For AI products, every state a result can be in: thinking, streaming, uncertain, failed and handed off.
What comes out
- User flows and wireframes
- High-fidelity screens
- A design system library
Step 03
Test
Prototypes go in front of real potential users before anything is built. The data decides, and we iterate when it proves us wrong.
- Prototype tests in Maze with real potential users.
- We define the ideal path, then measure perfect-flow completion, misclicks and abandonment.
- Usability interviews to hear why people struggled, not only where.
- A/B tests where two options are close, such as the two home structures we are testing for Allo Taxi.
What comes out
- Test results with recordings
- A list of changes, made before launch
- A validated prototype
Step 04
Build
We build it with our team or hand a system to your engineers that already covers every state, then measure after launch.
- A structured Figma file with design tokens and a component library.
- A clickable prototype of the intended flows.
- Optionally, a working front-end build with mock data, so developers see the real behaviour instead of imagining it.
- Production development, front end and back end, when you want us to build.
- Measurement after launch against the goals agreed at the start.
What comes out
- Developer-ready design files
- An interactive prototype or production code
- A post-launch review
What we hold to
Six rules behind every project
A product is a business before it is a screen.
Understanding how the product makes money is part of our intake. Design that ignores the business model does not survive.
Products convey feelings, not only functions.
We ask what the product should make people feel, and that answer shapes everything from structure to microcopy.
Process is a toolkit, not a ritual.
Design thinking works when it is not a straight line. We use the right method at the right moment, including going back when the evidence says so.
Build systems, not screens.
Every project gets a reusable design system, so the product stays consistent today and is cheap to extend tomorrow.
Be proven wrong before launch.
Testing exists to kill bad assumptions while they are still cheap to fix.
Getting started
How a project starts
- 01
Send us a message
Tell us about the product, where it is today and what you need. Start a project.
- 02
A call
We talk through the product, the goals and the constraints.
- 03
A scoped proposal
Scope, steps, timeline and price, for this project only.
FAQ
Questions about how we work
What is your UX design process?
Four steps: understand, design, test and build. We research with real users, design flows and an interface on a design system, test prototypes with real users before launch, then build or hand off and measure after launch.
Is the process always the same?
The four steps are, but the methods inside them change with the project. A redesign starts with an audit and data, a new product with interviews and competitor research, and an AI product with a review of the model's real output.
How long does a project take?
It depends on scope, and we agree the timeline on the first call. As a reference, MAP's marketplace and supplier dashboard were designed together in eight weeks, and Traveln.ai's map feature took a few weeks.
How do we start working together?
Send us a message about your product. We reply by email, set up a call, and follow with a scoped proposal.
What do we receive at the end?
A structured Figma file with tokens and components, a tested prototype, and, if you choose, a working front-end build or production code. Research findings and test results come with the evidence attached.
Tell us what you are building.
We reply by email to set up a call. Most projects start with a short conversation and a scoped proposal.