Design Without Limits: How AI Is Rewriting the Creative Process

As evidenced by Figma’s recent blockbuster IPO, it’s become abundantly clear yet again that there’s enduring value in platforms that elevate taste—especially when they do so through intuitive, collaborative workflows. But as we move deeper into the AI era, an important two-part question emerges:

What does it mean to have taste—and can we fully democratize it?

Taste is the ability to discern what is of good quality and high aesthetic standard.

Last fall, Silicon Valley found itself debating whether Taste is Eating the Valley, a counterpoint to the long-standing idea that Software is Eating the World. In a world saturated with software, design has become the differentiator.

What Is a Designer in the Age of AI?

Becoming a professional designer traditionally required a significant investment of time, money, and access. Prestigious institutions like RISD represent a certain model—respected, but expensive and exclusive. Mastery of tools like Adobe, Figma, Canva, or Blender demands training, mentorship, and years of experience. That barrier to entry made hiring a professional designer the only reliable way to achieve high-quality creative output—for branding, UI/UX, marketing materials, and more.

But that tradeoff is shifting. Today, AI-powered design tools are enabling a new class of creators to do work that once required formal expertise. Individuals can now produce high-quality assets at a fraction of the time and cost—with far more collaboration, speed, and iteration.

Platforms like ComfyUI and Flora simplify creative workflows through intuitive prompts and workflow editors. Vizcom makes 3D modeling accessible to non-specialists, while Arcade lets users design and sell physical goods with minimal friction. Meanwhile, tools like Lovart, Visual Electric, and Paper combine Figma-like design interfaces with real-time AI copilots to guide users from concept to completion—no professional designer required.

A few interesting highlights:

• Lovart’s Replay feature (see example) lets users scroll through the full prompt and output history, enabling them to revisit and explore any point in time or version.
• Visual Electric’s Merge tool allows users to “combine two images into one, applying elements from one image (textures, patterns, objects) directly onto another with a simple prompt.”

Rather than relying on static templates, users can generate a first draft and iterate on top of it. The bottleneck shifts from technical execution to idea quality. Good taste becomes a function of imagination—not access.

Here’s another example from Visual Electric, in addition to their homage to Pinterest inspo page. Using natural language alone, one can instantly iterate on top of a photo with fast inference time, high level of instruction following, and tasteful human-in-the-loop suggestions, all on a collaborative infinite canvas.

All this doesn’t eliminate the need for professional designers—it just changes when and why you hire them. Routine tasks will increasingly be automated. Professionals will focus on system-level thinking, aesthetic strategy, and high-concept work. The best platforms will bridge consumer accessibility with pro-level power, enabling creativity at scale.

From Slop to Taste: Building and Measuring Better Creative Tools

AI design tools face a core tension: Do they enable beautiful, tasteful outcomes—or just flood the world with more slop?

With one-click generation and auto-layouts, it’s easy to create something. It’s much harder to create something good. So the next question becomes: How do we build models and tools that elevate taste rather than dilute it?

It starts with measurement. Taste is subjective, but evaluative frameworks are emerging. General benchmarks like LMArena and Artificial Analysis already help assess model outputs for creativity, helpfulness, and correctness. Now, design-specific benchmarks like DesignArena are tackling the challenge of evaluating aesthetics.

Well-designed leaderboards could track:

• ‍Aesthetic Judgment: Which outputs exhibit stronger composition, balance, or fidelity to intent?‍
• Taste Alignment: Do the results reflect a particular brand, designer, or cultural moment?‍
• Usability-by-Default: Do tools nudge users toward layouts and systems that actually work?

These benchmarks are more than diagnostics—they're a forcing function for better models and better design agents.

Taste can also be embedded directly into tooling through two key methods:

1. Curation over Generation: Prioritize a constrained set of high-quality options, rather than endless output.
2. Feedback Loops: Offer subtle, actionable critiques based on best practices, helping users build their own intuition.

The best tools don’t just let people design—they make them feel like great designers. This emotional resonance is just as important as raw ability.

Rebuilding the Creative Stack from First Principles

Incumbents definitely aren’t ignoring this moment—Adobe’s Firefly suite is a clear sign of this competitive reaction. Form factor and user expectations are evolving rapidly, while legacy players are often weighed down by complexity and outdated UI assumptions. When Figma was founded in 2012, Sketch and Adobe XD were the large incumbents. Yet startups have a unique edge: they can rethink the creative stack from first principles. What if designing felt more like a conversation than configuring a control panel? What if we prioritized speed, delight, and accessibility over technical precision?

At Chemistry, we’re excited about this next frontier. A future where creativity is accessible, and taste is embedded in the tools we use. One where design is no longer the privilege of the few, but a fluency for the many. For those building these AI-native platforms: let’s raise the bar together—not just for what’s possible, but for what’s beautiful.