Two Non-Traditional Developers Share Their Journey Building Apps with AI

Vibe Coding Is empowering, but still requires user expertise.

Vibe coding is a rising trend in AI-assisted software development that allows users to build functional web apps, landing pages, and tools by simply describing their vision through natural language prompts. Instead of writing full codebases, users can converse with an LLM, giving it instructions and letting the tool generate the code. This approach is especially attractive to non-traditional or returning developers, creators, and entrepreneurs who want to prototype fast without getting bogged down in syntax or frameworks. But is it really viable beyond the MVP? To find out, we spoke with Monica Veggalam, a leadership coach and former Microsoft software engineer, and Monica Borrell, Director of Delivery at Olio Apps and a non-technical founder. They shared their firsthand experience building apps with AI tools like Bolt and OpenAI on the Code and Cognition Podcast.

---

Meet the Vibe Coders

Monica Veggalam (MV): She has recently revisited development using vibe coding to build landing pages and full-stack apps for her own use in her leadership coaching practice. Monica Borrell (MB): She experimented with vibe coding as a way to understand it by building a couple of personal projects.

---

Project 1: Landing Pages and Workshops

MV used Bolt to build a landing page for a leadership workshop. Traditionally, this would involve a designer, developer, and weeks of work. With vibe coding, she completed the task in two days. Her Key Takaway: Vibe coding excels at building static websites and landing pages quickly with good design and minimal code editing.

---

Project 2: "The Studio" : A Personal CRM for Coaches

MV also attempted a more complex app: a personal CRM to track coaching conversations, follow-ups, and ideas for writing topics and projects. While the AI-generated prototype was promising, bugs and UI issues emerged: Lesson Learned: Complex, stateful apps require progressive development and human debugging. AI has a tendency to introduce messy workarounds that make long-term maintenance difficult.

---

Project 3: Dream Journal with AI Analysis

MB created a survey engine application to act as a front end for Olio App's Data Pilot tool to streamline data enrichment and to extract insights from surveys. They did a little research that encouraged them not to try having the LLM build the application all at once and to break it into separate screens and functionality. However, once the application grew it couldn't add new or update functionality without breaking something first. However, scaling the app or adding new features eventually proved tricky: MB also created a dream journaling app that allows users to log their dreams and build the habit of dream journaling. It calls OpenAI for dream analysis. Lesson Learned: Keep MVPs focused. Vibe coding thrives when scoping is minimal and as complexity grows there is a larger chance the LLM will break your application.

MV's and MB's Best Practices for Vibe Coding

1. Break Down Features Start with one screen or function at a time. Building your app step by step or one screen at a time helps streamline the vibe coding experience.
2. Utilize a PRD Having a project reference document that describes the project being built, the tech stack you want to use, and other essential parameters of your project is helpful not just for the project owner, but can be utilized as a context document to your AI Code assistant to help keep it on track.
3. Expect Debugging Your AI is not going to create production ready code at the start. This generated code often requires human review at each stage of the project.
4. Vibe Coding is Great For Prototypes If you need to develop something to acquire and start testing for your first few users, vibe coding can be an excellent way to begin. However, it isn't sufficient for creating code capable of handling large scale deployment.
5. Keep Prompts Clear LLMs like ChatGPT and Claude, which Bolt uses under the hood, perform better with structured, detailed prompts.
6. Treat AI As a Junior Dev AI coding assistants are enthusiastic, fast, and can produce some messy code. This means as users we need to adopt the view of an architect to make sure the LLM is coding within the lines of an effective application.

What Tools Did They Use?

1. Bolt.new: A no-code/low-code tool powered by Claude AI.
2. Open AI: Home of ChatGPT and provider of LLM APIs

Other Tools Mentioned

1. Cursor: An AI-powered code editor.
2. Replit: Alternative to Bolt with a growing freelance debugging marketplace.
3. Devin:A GitHub-connected AI assistant for writing PRs and running tests.

Where is Vibe Coding Headed

Both Monicas believe the future of vibe coding lies in personalized apps, AI debugging assistants, and team workflows that wrap AI into conventional processes like pull requests and testing. MV is continuing work on her coaching CRM and content ideation tool, while MB wants to build a custom food tracking app tailored to her exact needs. If you'd like to listen to the full conversation, check out the episode on the Code and Cognition Podcast

Final Thoughts

After exploring the experiences of Vibe Coding with MV and MB, one thing is clear: the role of vibe coding isn't optimized for replacing developers outright but by extending the capability and creativity of anyone who has vision for a piece of software. Vibe Coding opens the door to new possibilities for solopreneurs, creatives, and returning coders, to build the applications they have always dreamed of, as well as the applications they think would be fun to try out. Ultimately, it opens the door to fast, expressive software prototyping. While it's not (yet) a replacement for production grade software development, it's an exciting new tool in the builder's toolkit.

Stay in Touch

Monica Veggalam: LinkedIn | Writes about leadership and personal growth Monica Borrell: LinkedIn | Director of Delivery at Olio Apps Want help debugging your vibe coding project? Reach out to Olio Apps for assistance in app development and production-quality support.