When Code Is Cheap, Design Becomes Your Moat

01 /The Concept

Ruby is Ion Design's AI product designer – and yes, she's a she. What matters is what she's not: not a prettification layer, and not a tool that simply polishes individual screens for usability.

Ruby does the full job of a product designer: mapping user journeys end to end, optimizing each screen for usability, and arriving at a polished visual result. She typically proposes multiple options – different approaches for different audience segments, or conceptually distinct alternatives.

The workflow runs in four phases:

- Ruby starts by studying the product's target audience and usage data – session recordings, for example.

- She then proposes new or improved user flows, presented as wireframes delivered directly to Slack.

- The team discusses the wireframes and offers feedback. Ruby incorporates that input and generates revised options.

- Once the team selects a direction, Ruby generates production code and opens a pull request – at which point developers can review and refine the details.

To understand what's expected of her, Ruby studies product documentation and can find reference examples online, though you can also provide them explicitly. Giving her a brand and design system from the start ensures new designs stay consistent with existing visual identity.

One notable detail: Ruby doesn't hand off designs to developers immediately. She first runs internal user simulations – emulating how real users would interact with the design – so developers receive pre-tested options for evaluation.

Ion Design went through Y Combinator in winter 2024 and just announced the launch of this new platform version on the YC website.

02 /Why It Matters

This is actually Ion Design's third platform – and each version has been a meaningfully different product, even as they've stayed within the shared theme of interface design.

The Y Combinator entry was an AI tool that converted Figma prototypes into React code, speeding up frontend development by roughly 40%.

Version two – [covered in April](/review/kak-rasshirit-uzkoe-gorlyshko-svoego-startapa) – was an interface editor and code generator that let product managers and designers build and iterate on interfaces independently, without pulling developers off their primary work.

Version three – Ruby – completes the arc: an AI designer capable of generating interfaces autonomously. With this version, companies could in principle reduce their design headcount while ending up with a more usable product, not a worse one.

What's textbook about Ion Design's trajectory is how their understanding of the core problem evolved with deeper domain immersion.

Phase one: build an AI tool to speed up what's most visible on the surface.

Phase two: realize that design and engineering are actually separate processes. The right move isn't "speed up in general" – it's to decouple the two so they can run in parallel. This isolates design as a distinct task within the original theme.

Phase three: optimize that isolated design process – and discover that the hardest part isn't converting designs into code (solvable). It's creating the designs in the first place.

The understanding of what the startup should actually be solving had completely changed. But it only became visible after the founders had built enough to see it – you can't identify the real problem from the outside.

This is the broader lesson: if your startup is still hammering on the problem you started with, there's a good chance it's not the most valuable one in the space. The topic might be right. But the specific problem worth solving usually only reveals itself after you've been working in the territory for a while.

03 /Opportunities

Before each batch, Y Combinator publishes its Requests for Startups – where YC partners describe areas they find most compelling right now.

The summer batch RFS included a section titled "More designer founders." That's a notable shift – YC has historically skewed toward technical founders. Here's a rough distillation of the argument.

In the coming years, AI coding platforms will accelerate software creation, resulting in far more software products competing for attention. Against that backdrop, design becomes the decisive differentiator – and some of the most successful YC alumni bear this out. Airbnb's design is arguably what made people comfortable renting their homes to strangers, and sharing payment details with an unknown platform.

Designers are understandably worried about being replaced by AI. But the more useful framing is what this opens up. At minimum, designers can use AI to dramatically accelerate their workflow. More importantly: since AI can now handle significant portions of coding, designers can build products themselves and start their own companies. In the not-too-distant future, "designer" and "founder" may be nearly synonymous.

AI coding platforms have already had their breakout moment. Cursor raised $900 million at a $9.9 billion valuation last month. Windsurf's founders turned down an acquisition offer from OpenAI and instead joined Google through a deal valued at $2.4 billion.

If design becomes the next key differentiator for software products – as the YC partner argues – then AI design platforms could climb to similar heights.

If you missed the AI coding platform wave, there may still be room to build the equivalent for design – whether that means letting developers ship without a designer, or letting a designer become a founder on their own.