Designing AI-native products from day one
Nad Chishi, founding designer at Lovable, is helping build one of Europe’s fastest-growing startups by fundamentally rethinking how design and product development should work when AI is at the core.
Instead of adapting legacy tools, Lovable is creating an entirely new model — one where LLMs and user experience are tightly interwoven from the ground up.
Start with small tools, not big visions
Nad began by hacking together LLM-powered design utilities: tools that checked color contrast, generated Tailwind color scales, and assisted with accessibility. What started as small experiments became a realization — AI could fundamentally change how designers work.
"I got completely bitten by this bug... this is what the future is going to be like"
Rather than build for where models might go, Nad emphasizes building for what models can do now. Lovable avoids being too visionary and instead focuses on rapid, scrappy iteration with short feedback loops.
Rethinking the design tool
The team originally built a familiar interface with layers panels and properties — then deleted it. They realized much of what’s considered “normal” in design tooling is just legacy.
"We built this sort of design tool version first, had an existential breakdown, deleted half of it, and then shipped a simpler version"
Their north star became flow and instant feedback — not matching the expectations of past tools. Features like visual edits emerged naturally, focused on balancing AI assistance with direct manipulation.
AI as a creative partner
Designers at Lovable don’t just prompt AI — they jam with it.
"Treat AI like a creative collaborator. Don’t expect perfection — just go with it"
Nad uses AI for everything from live prototyping to creative direction. He recommends feeding AI different art styles, film stills, or abstract visual references — not just UI keywords — to break out of old patterns and discover new expressions.
Designing interactions with AI, not just around it
Nad quickly realized designing the UI isn’t enough. You also have to design the system prompts, context windows, and model interactions.
"You realize setting the line height of the AI response isn’t really designing AI — what people do with it kind of is"
Lovable blurs the lines between design, engineering, and support. Anyone in the company — even non-engineers — can contribute to shaping how the model behaves.
Letting AI reshape the org chart
Nad argues that AI-native companies don’t need traditional silos. Many roles that sit between product and customers — like PMs, solutions architects, pre-sales — could collapse into fewer, more fluid functions.
"We can’t just edit the UX on one hand and think about AI on the other — we had to think about them completely together"
Building in flow
Lovable’s guiding philosophies:
Clock speed over perfection: Make fast decisions and learn through doing.
Small scopes first: Don’t overbuild. Iterate from the smallest version possible.
Flow as a product metric: If something enhances creative flow, it’s valuable — even if it’s technical.
The future: code as the new design source of truth
Nad believes the future of design tools will mirror developer tools: multiple interoperable environments built on a shared source of truth — not monolithic design files.
"Code can be the source of truth rather than design files"
This opens the door to a long tail of niche tools, focused on different domains like motion, shaders, or styling — made possible by LLMs and composable architectures like MCP.
Listen to the full episode here.
Source: Dive Club podcast
You might also like
