I vibe-coded an entire SaaS in a few months with Replit — here's what I learned

Hey IH 👋

I'm Dan — I run a coffee roasting business on the Gold Coast (Australia) doing ~$1.5-M/$2m year. I've been wanting to build software for years but never had the skills.

Then vibe coding happened.

What I built: AssetShark (assetsharkapp.com) — asset tracking and depreciation software for Australian small businesses. You add your business assets (laptops, vehicles, coffee machines, whatever) and it automatically calculates depreciation using official ATO rates. Spits out reports your accountant actually wants at tax time.

Why this problem? I've run small businesses for 20+ years. Every year at EOFY, I'm scrambling with spreadsheets trying to figure out what my assets are worth and how much depreciation to claim. Talked to other business owners — same story. The existing tools are either way too enterprise or way too expensive for a small business with 20-50 assets.

The stack:

• Built entirely in Replit. First with AI assistant (pre December 2025) and in the end using only their AI agent (it used to suck not it doesn't!)
• React frontend, Node backend
• No code written by hand — all prompted

What surprised me:

• Pre December 2025 it used to get stuff wrong all the time. From then on it pretty much nails everything. The change has been insane.
• How fast you can go from idea to working product (weeks, not months). This one took a while but since I built this I've done others in days. If I was to build this now it would probably take days.
• Marketing is still the hard part. Building the product was the easy bit. I only have a few users, time will tell if I'm able to turn it into a business or it remains a tool for me with a few people using it.
• I've been playing around with an AI tool (Viktor) to do SEO content, calculator tools, and content while I focus on the product. Again time will tell if that ends up being useful but it's certainly impressive.

Where I'm at:

• Product is live and working
• Free depreciation calculator getting organic traffic
• 24 SEO pages targeting long-tail depreciation keywords
• Blog with custom CSS doing weekly posts
• Looking for my first 50 beta users (4 so far)

The honest bit: Getting users is 10x harder than building the product. I've got 2 test users so far. If you run a small business in Australia (or know someone who does), I'd genuinely love feedback: assetsharkapp.com

If I was to do it again now I’d just use the agent and it would be a million times faster. I’d probably also build a mobile app which I’ve done with 2 other projects(one with Claude in terminal, and one with Replit) and it’s now very easy as well. When I built this, not so much.

Happy to answer any questions about the vibe coding process or the business side of things.