What you need to know about Vibe Coding as a Product Manager

We're still early; you don't need to become a vibe coder overnight. However, knowing about the tools and how they can fit into your workflow could give you and your team an edge.

AI coding tools are gaining in popularity, and if you've been anywhere near social media, you may have heard the term "vibe coding" recently. This concept is gaining traction and has already expanded to "vibe marketing", and isn't something you should ignore.

As a Product Manager, you should at least be aware of what vibe coding is, and although AI coding tools can play a valuable part in the development of your product, think of it as a turbo boost for now, not a replacement for developers. Don't expect everything to change dramatically for you and your teams, at least not yet.

With vibe coding, the idea is simple:

1. Describe what you want to build - just like writing a PRD.

2. Let the AI build it - No need to know how to code, but it can help.

3. Debug - Copy and paste any errors back to the LLM, instructing it to fix them.

4. Launch and iterate - Hit publish. Need to make changes? Just tell the AI and re-publish.

Does this mean you no longer need to work with your engineering team, instead doing all of the development yourself? Not at all. Besides, you have enough to do already, and there are some limitations with vibe coding preventing this from happening:

• Large code bases can quickly become unmanageable and non-human readable, making debugging difficult.

• Many existing production systems are proprietary and aren't documented well enough for AI tools to understand, limiting their effectiveness.

• You can sometimes get bad vibes and may need to start over if you get stuck in a dead-end loop.

• AI tools, especially UI-focused ones, often re-generate entire applications to make one small change, which is inefficient and messy if committing code to version control like Git.

• Security is still a huge responsibility that AI coding tools often overlook.

The Tools

There are two categories of tools. Low or no-code tools, and code first. Here's a short list of some of those tools, but there are more than what's listed here, and new tools are always popping up.

Low-code tools:

Coding is optional here. Visual editing and updating via prompting are the ideal modes of operating. Start from nothing, and speak something into existence. These tools include:

• Lovable

• V0

• Bolt

• Replit

• Rork

• Firebase

Code first tools:

You prompt the AI to generate code for you, but you also review the code, reference specific snippets in your prompt, define your own rules and work with an existing codebase. They require a code editor to work. Popular options are:

• Cursor

• Windsurf

• Copilot

• Cline

You might be wondering why I didn't mention Claude.ai or ChatGPT. Although they can write some code for you and even generate a preview, they are limited in their capabilities compared to the others unless paired with an agent like Cursor, at least for now.

How Vibe Coding Impacts Product Managers

You could proceed with business as usual, ignoring vibe coding for quite some time, but if you're here, you want to be ahead of the curve. So here are some ways you and your team can leverage and get the most out of these tools now:

• Create rough prototypes - Great for creating rough prototypes and refining new ideas before presenting with design or engineering teams; think of this as your wireframe mockup, but fully interactive.

• Speed up development - Your developers may already be using the code-first version of these tools, tab completions, and the ability to get rough scaffolding done easily can significantly speed up development.

• Micro tasks - Trying to "one-shot" an application rarely works, although it might give you a good starting point. From there, you should limit each request to a well-defined and scoped change you'd like to see. Grouping multiple changes in a single request leads to poorer results.

• Code mobility - It's fine to start in Lovable, Bolt or another tool of choice, then export your codebase, and make the final touches in Cursor. Get the help of a developer if you need to put the finishing touches on your prototype.

• Build internal tools - Internal tools are often neglected or non-existent. Many product managers do product operations work, and AI coding is a great way to help you build internal tools to automate or streamline team processes.

• Test user flow variations - It can be hard to imagine what a dramatic shift in how your product works would feel. Now, you can create multiple variations that are fully clickable and even functional. Leverage these for user testing and validation.

We're still early; you don't need to become a vibe coder overnight. However, knowing about the tools and how they can fit into your workflow could give you and your team an edge. If you're already experimenting and not happy with the results you're getting today, try again in a week or two; the speed at which these things improve is truly remarkable.