Grocery Prices are too (dang) high, can AI help?

Like many of us, in the last couple years, I've gone from casually shopping without worrying too much about prices to getting a lot more strategic with where I buy groceries and when. Because inflation. Because who wants to pay $10 for the same milk when you can get it for $6 elsewhere. So...I joined Costco, I started shopping at Trader Joes...and...I vibe coded to try to find a solution. I was also curious how easy or hard it would be to write my first form of vibe coded software, and comparing grocery prices seemed like a good place to start.

The Broader Context and Question:

Will people start building their own tools to solve household level problems? And will we see surge of apps and smaller tools built by individuals with an idea - now that the barrier to entry to go from idea to MVP is so much lower?

What I Set Out to Create:

Compare pricing across different online platforms on staples which I reorder frequently. Products like Viva Paper Towels, Oranges, Windex, and Luna Bars.

The Platform:

Claude, for it's reputation of being good at supporting vibe coding.

The Process

Here is the exact prompt I gave Claude after some initial discussion. I've found that the clearer a prompt is written, the better the initial results - like writing a brief for your robot.

With this prompt, Claude made me an interface that looked roughly like this:

Pretty cool, right? I am impressed it made this UI all on it's own.

The issue, and the main thing I had to overcome after this, is that Claude repeatedly refused to pull pricing. I asked Claude to scrape pricing from the websites, but it said it couldn't and kept asking me to enter in the prices myself (which, lol, takes away the whole purpose of the application).

Claude then suggested other options on how to pull price and complete this level of functionality. I love that Claude will suggest how to get things done, and provide options - this is super cool, and how we arrived on making a python script.

So, shockingly this worked. Claude walked me through how to download python, what to write on the command line, and eventually gave me a script to run that well, mostly worked. Here is what some of the output looked like:

User experience:

1) Enter in your products:

2) Approve your products

3) Check Prices!

4) Show the best deals:

I was excited to get this far. Of course to make this more than a novelty, it would need to have a better check-rate (it found 11/21 prices on products I approved), and ideally would run on it's own daily, notifying me about deals, versus me having to run the script. And it would get better at the "per unit" price calculations.

Insights:

• I am surprised no one has built a grocery comparison engine. I expect we're going to see a lot more of these household task type softwares being released. Moms may be the future of AI, and no one sees it coming :).

• Anyone who isn't scared can build a lot of core loops of a basic product, and that is very cool and kind of scary. That said, refining a product experience still will be a large part of the work to go from idea to actual business, or even thing you like to use for yourself. And, the barrier to entry on being scared to create products with vibe coding may still be kind of high. (for example, people hate to learn excel, that it's pretty easy to learn). I think we are both over and underestimating the level of friction here. Perhaps the future will belong to the curious, and those who have spare time.