Vibe coding as a non technical founder

I’m not a coder. That being said, I do work in tech as the CEO of a B2B SaaS and collaborate with developers daily.

I do know many tech concepts and even took some courses in html, css, js but also C++ and Python over the years. I’m curious.

Basically, I’m pretty bad at everything and wouldn’t be able to code something clean if my life depended on it. However, I’m not afraid of anything and can figure lots of things out if you give me some time.

It feels like I get stuck on issues that would be easy to solve as a professional developer… But at the same time, I don’t have a curse of knowledge and just try to get things done. Which sometimes means that I can get good enough results fast.

Here is my current point of view of Vibe Coding.

What I’ve done with vibe coding so far

• Deployed CRON tasks to sync a db from an API, daily
• Fixed typos automatically in Edusign
• Fixed lots of UI stuff in Edusign
• Built custom apps for Edusign’s app marketplace, with external APIs and UI
• Launched small marketing websites/experiments
• Built a SaaS (Digital Asset Manager), working in production with auth and full features.

Tools I’ve tried

• Replit <3
• Claude Code
• Codex
• Lovable
• V0
• Figma Make
• Dust
• Copilot
• N8N, Make, Zapier
• Perplexity/Claude/ChatGPT
• And many smaller ones for specific use cases I’m forgetting.

Things I’ve learned

• Plan as much as possible
• Split into small digestible bits (like a human!)
• Treat it like a kid with super powerful powers
• How to use Git & Github, way better
• Many tech concepts
• Search what you don’t know
• Shoot your shot but don’t persist, don’t be that lazy and read some code

What I like about it

• Feels like magic
• Can focus on the UX, finally
• The UI is surprinsingly good
• Spent 80h vibe coding in a week, wanted to do even more

Where I struggle

• Some things can break
• Basics are tiring sometimes
• Software is still hard

What I wish for the future

• Better sandbox/production segmentation, but easy to go from one to the other.
• Smarter software around the LLM. I love what Replit is doing with some contextual buttons for example, way better than average in my opinion.
• Obviously, better reasoning models

What I shared with my team

• Cultivate curiosity & fundational skills. It won’t be about typing the right thing, but what’s the right « way » to do it.
• Great experience designers are going to have a great time. Average one aren’t going to be there to see it.
• Use AI today to be ready for tomorrow. Because it’s coming no matter what you wish.
• Claude Code is cheap compared to your salaries, stop comparing company expenses to your personal money.
• If only 10% of the time you 10x you could go twice as fast. Yes it’s going to be wrong, often. But when it works… It’s a marvel.
• For the non technical people ; Use AI, and learn basic tech concepts as well. If you don’t know what’s Github, a webhook, a db or an API you are going to have more surprises than you might expect.

Controversial take

I do think it’s possible to launch a SaaS 95% Vibe coded and have some users use it. But not just as an experiment or a PoC. A real fully working SaaS that won’t break too soon either.

But why are all the posts on Linkedin, X or Reddit saying otherwise?

• Non-tech people lack some technical concept to reach that possibility in the majority of cases
• Easier for devs to shame people on this topic than to support the growth
• The few that do it successfully don’t share it because of the previous points
• Like every other new business, only 1% really make it, the rest is noise.

So what does it all mean?

Well, I’m not 100% sure to be honest.

However, I know for a fact that even if Vibe Coding doesn’t end up being the way to code in the future, it will help the world on one very important point:

Coding is no longer scary.

This is the biggest deal. It’s not just true for coding by the way. The greatest perk of LLMs, in my opinion, for the average human, if used correctly, is to be able to learn faster.