Why Product Managers Need to Embrace Vibe Coding

Background 

Over the past few weeks, I have been building various apps and products I have always wanted to create, but lacked the skills and financial means to do so. While cleaning up my Google Drive, I stumbled upon a few files containing detailed user stories and simple app ideas I had developed a long time ago. Today, after much waiting, I get to finally vibe code these apps. Looking back a few years, building these apps seemed impossible. If it still does, check out my blog on how to start with prompt engineering.

One thing to keep in mind is that change always brings some level of discomfort. There was this weird feeling I had as I was vibe coding, and I started to wonder if I understood Co-Pilot’s outputs. Fortunately, I have worn many hats in my career. I have worked as a developer and product manager. As I was vibe coding, I wasn’t sure what hat I had on. Was I a developer? Or was I still a product manager? I am a developer, but I am far from being a skilled one. When I look at the code in my IDE, some of it never made any sense. That is when I realized that what I can do is describe a product very well, so a large language model (LLM) like Claude or Gemini can understand what to build. In this blog, I want to highlight how vibe coding is changing product management.

1: Time to Show and Tell 

As a product manager, you rely heavily on data to improve your products. You can see which features work best for your customers and identify any new features you need to add, all driven by data. In the past, you would create a Product Requirement Document (PRD) or a lengthy list of user stories. But now, with Vibe coding, you can build a quick, no-code prototype and explain to your team what you want to make next. A prototype can save you a significant amount of time and energy spent trying to explain your vision. If you choose to build the prototype using tools like Co-pilot or Cursor, you could save your development team some energy, allowing them to continue optimizing the code.

2: Security first approach

While building Link Speghatti, I made deliberate decisions about the information I provided in response to the prompts. I didn’t want to accidentally provide my AWS keys or Stripe keys. The security-first approach can save your company from AI agents that not only understand your code better but can also help prevent bad actors from misusing your information. While doing this, I also learned about environment keys and similar to my previous post, my appreciation for the developers and all their effort grew.

3: Different between real and hallucinations 

Yes, LLMs can hallucinate. They can take you down a rabbit hole and suggest features that can seem enticing, but add no value to your customers. As a product manager, it’s beneficial to brainstorm with an LLM, especially when asking questions about innovating new features, as it tends to forget the data you provided and starts to hallucinate features. Additionally, there is concern about how the LLM will utilize users’ data, which could potentially lead to a data breach.

4: Data Analysis is Faster

Suppose your product is collecting data on user actions. In that case, you, as a product manager, can easily ask LLMs to analyze and create beautiful charts to understand your user behaviour. With this new power, a product manager can highlight trends, create charts, or even point out anomalies. With the help of LLMs, you can now be your own junior data analyst and make product decisions faster without relying on anyone.

5: Start Understanding Security Standards

When I first began building Nanny Bot, I didn't think about security. I just wanted to create. The more I dove in, the more I grasped the responsibility we have as product managers, even with AI handling the code. I instructed Claude to integrate with Agno to connect to my data source and save user chat data. The solution provided by Co-pilot was technically correct, but it wasn’t the right approach. The model suggested saving sensitive information in plain text. Considering the total lack of security, I need to jump in and start suggesting encoding standards.

Conclusion

With Vibe Coding, the product manager’s role is now turning into a show-and-tell. With the help of AI tools, you can build quick prototypes or analyze data whenever you want. As a product manager, if you begin vibe coding, you can save your team a lot of resources and time. Although AI tools can quickly generate code, they don’t adhere to security regulations unless you specifically instruct them to do so. Even then, they may invent solutions that appear reasonable but are entirely incorrect. As a product manager, you have the option to build quick prototypes to share your vision with your development teams and upper management. However, you need to brush up on your technical terminology, security standards, and understand what data you can provide to LLMs to prevent data breaches.

If you enjoyed this post, here are a few ways to go deeper:

• Check out my Prompt Engineering Guide: Prompt Engineering Playbook
• Connect with me on LinkedIn
• Read more on my blog
• Try what I built: Visit Bit Mentors Inc.

Let me know what you're building, happy to discuss ideas!