If you're a non technical founder trying to vibe code your idea into something real, this one's for you.

@sahanaintech

7.4K followers · usually 1.8K views a post

Views

13.2K

Likes

962

Comments

132

Selling · Course / Info

Sahana In Tech Notion Doc

https://sahanaintech.myflodesk.com/home ↗

Why it worked

The video addresses a common pain point for non-technical founders: the perceived barrier to entry in coding and product development. By offering practical, actionable prompts and demonstrating how AI can democratize technical tasks, it provides immediate value and positions the creator as a helpful resource. The "soft sell" approach, where the product (Notion doc) is presented as a natural extension of the valuable content, makes it more likely to convert viewers.

Summary

This video provides five prompts for non-technical founders to leverage AI for coding assistance, covering planning, automated tests, Git setup, security vulnerabilities, and code review. The creator emphasizes that these prompts help bridge the gap between non-technical founders and engineering best practices, offering a comprehensive Notion doc with more examples in her bio.

Structure

1Introduction to the problem for non-technical founders
2Addressing engineer skepticism and providing solutions
3Prompt 1: Planning and Tech Stack
4Prompt 2: Automated Tests
5Prompt 3: Git Usage
6

Original caption

Some engineers are just so smug with their “you can’t do this because you’re not an engineer” narrative. I just don’t buy it. So here are some ways you can apply the same principles senior engineers do while you vibe code. Specifically for people who are non-technical. Link to the full guide with 30 prompt examples is in my bio, it’s free. Save for later and let me know if you give it a try! #vibecoding #claudecode #cursor #nontechnicalfounder #founder

More from @sahanaintech

Their other posts in the index, biggest breakout first.

On-screen text

If you're a non technical founder trying to vibe code your idea into something real, this one's for you. I've been seeing so many comments from engineers saying "oh if you aren't technical you're vibe coding garbage" comments like this. And this. Come on. I'm an engineer. I've spent over 100 hours vibe coding and I think you, non technical founder, can get pretty far. but you need to know exactly what to prompt the LLM to have it act like a senior engineer. So here are four prompts right off the bat. I have so many more but I'll share a resource later. First, you need a plan. I'm talking a PRD and a doc that articulates exactly what technologies you wanna use. How do you get that? You ask the LLM. Prompt the AI to set things up so it always checks the PRD and the tech stack doc before making any changes. Prompt example: Given this PRD, recommend a beginner-friendly tech stack with reasons. It should have fast hosting, a built-in database, authentication, and minimal setup. Follow-up: Explain what each piece of this stack does in simple terms and what problems it solves. Save this in a tech stack document. Use this in every future prompt: Update CLAUDE.md / Agents.md with some rules to ensure every future prompt checks the PRD and tech stack doc before generating code. Add an instruction to emphasize modularity (multiple files) and discourage a monolith (one giant file). Before you introduce new technologies or packages always review if there is a currently technology I could use first. If not, add the package and also add it to this doc. #2 Prompt the AI to add automated tests. Writing tests is the primary way that engineers make sure that the stuff they're building actually works. It also helps you catch when the AI gets overzealous and changes a bunch of old stuff that you didn't want it to change. Ask the model to reason like a reviewer before merging code. Adding tests will ensure that future feature work doesn't break stuff you've already built. Prompt examples: Run a code review for this feature. Focus on readability, edge cases, and security. Make any changes you recommend. List potential failure points or bugs. Write tests that cover happy paths and edge cases. Explain what these tests verify and what remains untested. Read through the test cases that the AI generated, do they make sense? You should feel confident that if these tests pass then your app will work. You can use a prompt like this. #3 Learn how to use git This one's a little technical, but stay with me. Learn how to use git This singular tool will help you make sure that you always have older versions of your project available Maybe I'll make a whole another video on how to use Git. The good news is your AI tool can help you get Git set up. Use a prompt like this. Next (don't save things that are autogenerated, they don't need to be saved on Github): Add a .gitignore file for this stack that excludes build files, dependencies, and environment variables. Once everything looks right: Help me initialize a new Git repository and push this code to GitHub. Include the git commands step-by-step so I can copy and paste them. Once done, make sure you're inside your new repository in your coding tool. (Not sure? Ask AI) Use a prompt like this. #4 Prompt to check for security vulnerabilities. Don't be this guy that publicly shared his API keys. You can use a prompt like this. #5 Prompt to review the code like a senior engineer would. Self-review: Review this code like a senior engineer. What would you improve for clarity, performance, or maintainability? Make sure to format everything using Prettier and apply linting with ESLint using recommended defaults for this stack. You can use a prompt like this. I have so many more examples so I just put everything in a Notion doc. It's a step by step guide with a ton of prompt examples so you can do your best to build with engineering best practices in mind. My name is Sahana. I'm an engineering leader and I talk about tech, engineering and leadership. Have a good one!

Transcript