A Product Manager’s guide to rapid prototyping with Cursor & Claude Code
What every PM needs to know about building faster with AI pair-programmers.
Why realistic prototypes are relevant for PMs
As a Product Manager, there’s never a shortage of ideas for new features or even entirely new products. Users share the capabilities they wish they had, prospective customers describe problems they need solved, and product metrics reveal fresh insights into feature usage almost every week. The challenge isn’t finding ideas. It’s turning them into something real.
For non-developers like most PMs, creating a working prototype used to mean long waits and high costs. You’d write up requirements, hand them over to an engineering team, and then spend days, sometimes weeks, clarifying details, reviewing half-finished artifacts, and waiting for changes. This process was often too slow and too expensive to test whether an idea had potential. Yet the whole point of product discovery is rapid experimentation: quickly and affordably assessing the potential value of an idea, a feature, or a product. In my experience, the best way to sense that value is with something the user can see, use, and play around with. The goal is a fast, lightweight piece of software that allows you to experience first-hand which impressions it creates and which problems it reveals. Only then can you truly understand what resonates, what confuses, and what needs to be improved.
Today, modern AI coding tools like Cursor and Claude Code change the speed at which software can be built completely. They enable PMs and other non-developers to create functional, polished prototypes at a fraction of the time and cost. In other words, enabling exactly what prototyping was meant to be: a way to validate whether your solution meets the customer’s need or solves their problem as fast as possible. In this article, I will share how I use these tools for kickstarting my start-up Prodcovery, and the setup that helps me validate and test ideas quickly.
My setup to write functional code
Cursor
Cursor is essentially VS Code with AI superpowers. It lets you explore entire repositories, jump between files, and search your codebase just like the editor you already know, but with built-in coding agents to help along the way.
What makes Cursor stand out is its flexibility. It supports top models from OpenAI (GPT-5), Anthropic (Claude Opus 4.1 and Sonnet 4), and Google (Gemini 2.5 Pro), and it integrates smoothly with GitHub, Docker, and many other developer tools through its marketplace. Pricing is straightforward: there’s a free plan (where Cursor picks the model for you), a Pro plan at $20/month, and an Ultra plan at $200/month with more tokens and full model selection.
Claude Code
Claude Code is Anthropic’s dedicated AI coding environment, built to bring the power of its Claude models into the software development workflow. Instead of functioning like a traditional IDE, Claude Code is offered as a lightweight CLI tool that emphasizes natural language interaction with your codebase. It excels at producing clean, well-structured code, and for more complex tasks, its Plan mode asks the model to create a step-by-step execution strategy before writing or refactoring code. While it doesn’t provide the rich plugin ecosystem of Cursor, Claude Code is widely regarded as one of the best coding agents available. Developers praise it for its clarity, conversational depth, and the ability to work with very large context windows. The service is free to try with limited usage, with higher-capacity tiers available through Anthropic’s paid subscription plans.
Getting the best of both worlds
For my own workflow, I’ve found that using Claude Code inside Cursor gives me the best of both worlds.
On the one hand, Claude Code is simply the strongest coding agent currently available and Anthropic’s latest models underline this position. I use Claude Code in Pro mode, where you get larger context windows and smarter step-by-step planning capabilities. The Pro plan also refreshes its token allowance every few hours, which means you can run multiple in-depth coding sessions throughout the day without worrying about hitting a hard cap. For bigger tasks, you may sometimes need to wait a few minutes, or even hours, for your token limit to refresh.
As a valuable addition, Cursor provides the full IDE experience: I frequently open and edit files manually and take advantage of built-in Git integration for version control. Beyond that, Cursor’s agent mode also allows me to generate code which is very useful for conducting smaller, isolated changes while waiting for Claude Code to being fully-loaded with tokens again. The combination of both tools is very powerful—Claude Code handles the heavy lifting of reasoning about frameworks, architecture and generates high-quality code, while Cursor anchors everything in a familiar development environment where I can test, refine, and ship projects. Together, they turn what used to feel like juggling multiple tools into a single, cohesive coding workflow.
Conclusion
AI coding tools like Cursor and Claude Code aren’t just about writing code faster. They’re changing how non-developers, like most Product Managers (myself included), actually work. What used to take weeks of coordination with engineers can now be done in a couple of hours, in a form users can touch and react to. And that shift is huge! Prototyping is no longer a bottleneck but a natural part of everyday product discovery.
Of course, take this with a grain of salt. A prototype is not a production-ready software product. While vibe-coding enables people like us to create functional prototypes, it doesn’t replace the years of experience, context, and taste of a senior software developer.
For me, though, these tools have transformed how I work: they help me turn abstract ideas into concrete experiments I can put in front of users. And it’s not just about speed (though it is also about speed!)—it’s about learning faster, reducing risk, and having greater confidence that what we create actually resonates with our customers.
At Prodcovery, that ability to turn “what if” into “let’s try it” is invaluable. I believe this is where product management — especially product discovery — is heading. Stay tuned for my next article, where I’ll share some tips and tricks in Claude Code that make working with it even more productive and fun.
Thanks for reading and greets from Munich,
Christian