Most founders have sat through design reviews that felt like death by PowerPoint. Static screens, fragmented feedback across emails and docs, then hours of follow-up work for designers. It's a familiar drag.

Owen Williams, a design manager at Stripe, saw this problem firsthand. He decided to tackle it by building Protodash, an internal AI-powered prototyping tool. Its "design review mode" isn't just about collecting comments; it's about turning feedback into automated action. As Williams puts it, "being in a design review where I can click things is my favorite." This shifts the entire culture towards "demos not memos," a philosophy championed by fellow Stripe design leader Dan Nelson.

Instead of screenshots dumped in Google Docs, Protodash lets stakeholders comment directly on interactive, URL-shareable prototypes. But the real game-changer is what happens next: AI summarizes the feedback, queues up suggested fixes, and even applies those changes to the prototype automatically. Williams explains, “you get like a detailed summary of the design review at the top. And then you could actually enter like review mode and step through and be like, the filter pattern isn't right here. Please add three more options... and then that will add it to a queue that you can just send to the AI to fix.”

This drastically cuts down the post-review "busy work." Designers no longer chase down every minor tweak; the AI does it. Williams highlights the relief: "designers have to follow up. Like there's all this extra busy work that you have to do after a design review and being able to be like, I fixed that. Here's the receipts thread is amazing."

Key Takeaways

  • Stripe's internal tool, Protodash, directly addresses the inefficiencies of traditional design reviews, moving from static presentations to interactive prototypes.
  • Protodash Studio's "design review mode" enables URL-shareable prototypes where stakeholders provide direct, contextual feedback.
  • AI automatically summarizes discussion points, generates a queue of suggested fixes, and applies design changes to the prototype without manual designer intervention.
  • This automation significantly reduces post-review "busy work" for designers, shifting focus to strategic discussion rather than tactical execution of minor tweaks.
  • Owen Williams' approach operationalizes the "demos not memos" philosophy through Protodash's AI-Powered Design Review Method.

The Protodash's AI-Powered Design Review Method

  • Enter Review Mode: You can jump into a comment section. You can click start review and then like share a URL.
  • Collaborative Feedback Collection: everybody can comment as like a normal human
  • AI Summary Generation: you get like a detailed summary of the design review at the top.
  • Feedback Queue Management: you could actually enter like review mode and step through and be like, the filter pattern isn't right here. Please add three more options... and then that will add it to a queue.
  • AI-Powered Fix Execution: The accumulated feedback in the queue can be 'just send to the AI to fix like straight off the back of the design review'.
  • Automated Follow-up: The AI can then report, 'I fixed Katie D's feedback for you and you can send this to her'.

When This Works (and When It Doesn't)

This method transforms design reviews from 'drowning in presentations' and 'dumping ground of like screenshots' in Google Docs to an interactive, efficient process where 'designers have to follow up' is reduced, enabling 'demos not memos' and allowing 'the actual work' to be discussed. It shines in environments with established design systems and component libraries, like Stripe, where UI elements are well-defined. The AI can reliably interpret and apply changes based on these clear constraints.

However, this approach isn't a silver bullet for every design challenge. It falters in highly conceptual, early-stage design where ambiguity is high, or for nuanced brand-identity work that requires a human eye for subjective judgment. If your product lacks a robust, consistent design system, the AI might struggle to make intelligent, context-aware adjustments. Its effectiveness relies heavily on the underlying structure and predictability of the design elements it's manipulating.

What to Do With This

This week, stop tolerating the endless design review feedback loop. Identify one recurring, tedious design fix in your product development – perhaps adding options to a dropdown, resizing an element within fixed parameters, or adjusting component spacing. Don't aim to build a full Protodash. Instead, create a minimal proof-of-concept: a small internal script or tool that uses AI to automatically execute that single, repetitive design change. The goal is to get that first AI-powered win that eliminates a sliver of manual design labor, freeing your team for higher-level work.