Deep: How Stripe builds prototypes in the browser

2026-02-09 · Source: Department of Product · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Project & Product Management · Depth: Intermediate, medium

Summary

The concept of "vibe coding" is democratizing internal tool development within leading tech companies, allowing non-engineers like designers, product managers, and researchers to build useful applications. Examples include Duolingo's internal tool for product managers to create feature flag agents, Anthropic's mini-apps for lawyers, and Stripe's Protodash, an AI-powered prototyping tool developed by design manager Owen Williams. Protodash enables rapid, in-browser UX exploration using real data and Stripe's actual components, facilitating interactive prototypes over static Figma files. This Deep Dive will further explore Protodash's functionality and impact, alongside other internal tools like a data agent for 600 petabytes of company data, experimental UIs for custom dashboards, and an AI project manager coordinating dozens of agents across various Google services.

Key takeaway

For product managers and designers seeking to accelerate prototyping and internal tool development, embracing "vibe coding" and AI-powered tools like Protodash can significantly reduce friction. You should explore building lightweight, local applications that integrate your company's component libraries, allowing for interactive UX exploration with real data. This approach enables your teams to bring functional prototypes, rather than static designs, to meetings, fostering more dynamic and realistic feedback cycles.

Key insights

Vibe coding empowers non-engineers to build internal tools, democratizing development and accelerating prototyping with AI.

Principles

• Internal tools can serve as testing grounds for external products.
• Disposable, lightweight tools reduce maintenance overhead.
• Interactive prototypes enhance design discussions over static files.

Method

To build a Protodash-like tool, create a local web app mimicking your product, integrate your company's component library, and add an AI assistant for real-time HTML element selection and editing.

In practice

• Use AI assistants to generate prototypes with real components.
• Build local, disposable tools to explore UX ideas quickly.
• Employ "reverse prompting" to accelerate AI-assisted code generation.

Topics

• Vibe Coding
• AI-powered Prototyping
• Internal Tools Development
• UX Design
• Claude Code

Best for: Product Manager, Product Designer, Research Scientist

Related on AIssential

• 😺 🎙️ Learn Vibe Coding from Microsoft Build, LIVE — AI-powered software development for beginners requires understanding core tools and foundational concepts.
• Vibe coding is coming to your phone — AI coding tools are enabling non-developers to create personalized mobile apps, widgets, and automations, shifting towards generative UIs.
• A new worst coder has entered the chat: vibe coding without code knowledge — Vibe coding tools enable rapid app creation but often produce insecure, unmaintainable code without developer oversight.
• I Tested The Top 3 AIs for Vibe Coding (Shocking Winner) — Vibe Coding enables rapid, no-code website development using AI, emphasizing iterative prompting and debugging.
• Can a newbie really vibe code an app? I tried Cursor and Replit to find out — AI coding tools simplify initial setup but require programming knowledge and significant refinement for complex applications.
• #343 Vibe Coding and the Rise of the Non-Developer Builder with Matt Palmer, Developer Relations at Replit — AI-driven "vibe coding" empowers non-developers to rapidly build and deploy custom applications, accelerating product delivery.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Department of Product.