How I created an entire project using only Github Copilot and no-code tools

Projects / By

TL;DR: I’ve been coding for the last 15 years. I got access to Github Copilot recently. I wanted to see if it’s possible to build a functional web app without writing any code. I’ve used Github Copilot to create the code for the backend and no-code tools for the frontend.

Results: It’s possible. It’s a pain in the ass. It’s not good code by any standards unless you rewrite it all. You still need to know how to code.

The website

The idea

  • Create the most generic product/SaaS/web-app in an extremely saturated niche
  • Build it in under 2 weeks while spending less than $200
  • One-person online business
  • Launch an MVP as soon as possible
  • Promote it using social media and start making money rapidly
  • Sell within a year

What? Create a simple website that displays items from Amazon that are currently discounted.

Why? Most of the deals that I find online are either already sold out or the price doesn’t reflect what’s up on Amazon anymore.

So how do we do it? I only want to see the really good deals. And I want the price and availability to be rechecked using the Amazon Advertising API every 10 minutes max. (I got it down to 5-minute intervals for the time being)

So a few months ago I created my own bot coded in Python that was monitoring Amazon. After a satisfying deal was found it sent me a notification using Telegram. I shared it with my friends and got some positive responses.

Last week when I got the invite to Github Copilot I thought it would be fun to recreate that backend entirely from scratch and add a no-code frontend. Without coding anything by hand.

At this point, I didn’t realize what a crazy ride I signed up for.

Why?

I want to showcase several issues:

  • Founders waste too much time coming up with ideas
  • The product is often niched down too much
  • Development takes too long before an MVP is launched
  • Too much money is wasted on development before launch that could go to product validation and marketing
  • People don’t understand how to successfully promote on social media

The product

Took me an hour to come up with this:

  • DealsViking.com domain name was available
  • A website for cheapos like myself that finds deals and discounts on Amazon using their API
  • Monetize using Amazon affiliate program
  • Low-code/no-code development
  • Marketing focused around building in public, showcasing differentiating features, niche tech stack, sharing deal opportunities, and a newsletter referral program

The tech stack

Backend

Frontend

Why 2 databases? Softr is basically an Airtable wrapper. The sqlite3 database already has hundreds of thousands of Amazon products in it, while I’m keeping around 400-600 live deals in the Airtable.

What it cost me

  • $29/mo for Softr
  • $9/mo for Integromat, but I’m also using it for other stuff.
  • $49 for SendFox (5k email list) one-time payment
  • $0/mo for Airtable. The free tier supports 1200 records and that was enough for my project.
  • $8.50 for the domain name
  • My sanity /s

What’s wrong with Github Copilot?

I’m not going to comment on the copyright problems, as this has been already extensively covered by others.

  • It’s going to make you a worse programmer. Using Copilot is like copy-pasting from Stack Overflow but way, way worse. The code that you find on Stack Overflow usually isn’t 100% fitted to your needs and you have to alter it and understand it at least a bit.
  • Reading code takes more time than writing it. In my experience, I wasted more time trying to understand what Copilot coded and fixing it than I would coding it myself.
  • It’s terribly inconsistent. It’s using random programming styles, and doesn’t take into really account the other parts of your codebase. It produces security nightmares like, in my case, obvious SQL injection vulnerabilities.
  • It still outputs typos and random strings. Again, have to regenerate the code and waste more time.
  • The code is pretty terrible:

Marketing

I did basic things, but I worked very hard for a week.

  • I contacted all my friends, family, work network to let them know what I’m doing.
  • I started over 100 discussions in relevant subreddits, groups, forums, Indie Hackers, etc. I didn’t do anything black hat, spam, or any vote manipulation.
  • I tweeted a lot and let people who are into deals know that I’m launching.
  • I outreached to bloggers who feature deals sites and got added to a dozen articles.
  • I’m hoping the rest will be a combination of auto-tweeting of deals, word of mouth, and newsletter referrals.

1-week-after-launch update!

Did a solid week of marketing before the launch and the results were decent:

Conversion sucks but it’s to be expected with a service like this combined purely with social media promotion.

Also got 326 email subscribers with an open rate of 29.71%.

3-month update!

As of now, the project is entirely automated. I just take about an hour/week to look at the data and make sure everything is working correctly.

  • The project has been more successful than I expected
  • I gathered up some testimonials
  • We’re up to 2081 email subscribers as of 10/22/2021
  • Monthly profits crossed $3k
  • I got a 40k buyout offer last week, but decided to keep the site for now.
en_USEnglish
de_DEGerman en_USEnglish