Which Chatbot Is Best for Web Design?

(A PM Discovers the Real Problem. “Reader, it’s me.”)

Summary: In which I ask the internet which chatbot is “best at web design,” gently roast myself, and accidentally clarify a much more useful operating model for AI-assisted design, prototyping, and commerce UX.

The Question I Asked (Out Loud, Online)

“What is the best chatbot for web design and style? Claude kinda sux.”

This is the professional equivalent of standing in a hardware store and asking which hammer is best for vibes. Understandable. Relatable. Incomplete.

What I Learned (Against My Will)

There is no single “best” chatbot for web design. There is a best division of labor, and I wasn’t using one.

Claude doesn’t suck. Claude wants a PRD before committing to line-height.

The Actual Fix: Better Framing, Not a Better Bot

Once I stopped asking for “make it pretty” and started asking for decisions under constraints, the output improved dramatically.

The winning pattern:

This is not micromanaging the AI. This is product management.

Plot Twist: I Haven’t Used Figma AI Prototyping Yet

Yes, I know. I have strong opinions about design systems and had not yet seriously explored Figma’s AI-assisted prototyping tools.

This turns out to be a missed opportunity.

Used correctly, Figma AI can:

The goal is not pixel perfection. The goal is faster conviction.

Meanwhile, Shopify Is Quietly Shipping AI Hero Capabilities

While I was interrogating chatbots, Shopify rolled out AI-assisted hero banner tooling that can:

Important caveat:

These tools accelerate good systems. They do not fix bad ones.

AI hero generation works best when:

AI won’t save weak hierarchy. It will save you from opening Photoshop at midnight.

The Embarrassingly Obvious Conclusion

I didn’t need a better chatbot. I needed:

This isn’t failure. This is how new tools get domesticated.

Final Self-Assessment

Asking “which chatbot is best at web design” is extremely on-brand for a PM who knows the system, knows the taste, and briefly forgot to specify either.

We ship. We learn. We add constraints. Then we ship again — calmer, cleaner, and with fewer existential questions.