[ Design with AI ]
AI-native design for SaaS and AI products
We design with AI in the loop: more directions explored, prototypes in code, design that matches your codebase - and senior designers who decide what’s worth shipping.
[ Works with ]
Claude Code
Coded prototypes and design QA
Figma MCP
Design and code read the same tokens
Cursor
Fast edits inside your codebase
GitHub
Handoff where your developers work
Read the ticket and order history0.8s
Checked refund policy §31.2s
Drafted a refund of €240 to the original payment0.6s
Needs approval - refunds over €200 need a personnow
Refund €240 to the original payment method?
[ Why it matters ]
Screens are cheap now. Judgement isn’t. AI lets us explore wider and reach a working prototype sooner. People decide what fits your users, your product and your business.
[ How we use AI ]
Where AI earns its place
More directions
Agents help us explore several directions in the time one used to take. You choose from real options, not one guess.
Prototypes in code
Clickable flows on real components and data - something you can test with users, not just look at.
Design that matches your code
Tokens and components stay in sync with your codebase, so developers build what was designed.
[ How we work ]
People decide. Agents do the legwork.
Explore
Agents draft directions, patterns and edge cases. We choose what fits your users and your roadmap.
Prototype
Agents turn the chosen flows into a coded prototype with your tokens. We design the states and review every screen.
Hand off
Agents keep components and tokens in sync with your code. We run design QA on the real build.
[ Quality ]
Every screen has an owner
AI speeds up the work. It doesn’t sign it off.
- A senior designer owns each flow from start to finish
- A design director reviews every screen
- Product logic, edge cases and quality are decided by people
- Tools that touch your files are agreed with you upfront
- Drafts and prototypes
- Agents, reviewed by us
- Sign-off
- Design director
[ AI products ]
We design AI products, too
Chat, copilot and agent interfaces
Assistants people understand, check and control
Clear states
Running, done, failed or waiting for review - visible at a glance
Review and approval steps
People approve what costs money or can’t be undone
Patterns your team can reuse
A small system of AI patterns, not one-off screens
[ Testimonials ]
What people say
[ FAQ ]
AI questions we get
Every day. We use it to explore more directions, draft content and edge cases, build coded prototypes and keep design in sync with code. A senior designer owns each flow, and a design director reviews every screen.
Figma with Figma MCP, Claude Code and Cursor for coded prototypes, GitHub for handoff. We adapt to your stack and agree upfront which tools may touch your files.
Not for your product. We build coded prototypes, tokens and components your developers can reuse - production code stays with your team. Marketing sites we can design and build end to end.
Your files stay under NDA. We use only the AI tools you approve, and anything you mark as sensitive we handle without AI.
It makes exploration and production faster, so more of the budget goes to decisions, testing and quality - not to pushing pixels.
Yes - chat, copilot and agent interfaces, clear states, review and approval steps, and a small set of AI patterns your team can reuse. See Dynamiq, Quillora, ConvoGrid.ai and AIMFINA.














