Building with AI Before Breakfast: The Amazon Spending Analyzer

Instead of simply eating breakfast, AI with Breakfast began as a weekly ritual where I visit a café in Abu Dhabi, enjoy something good to eat and talk about two things I love: food and AI. That continues, but I also want to expand the idea of the substack.

On another day each week, I skip the café breakfast and use that same hour as a focused building session. The challenge is simple. In the time it would normally take to eat breakfast, I use AI to build a small tool or experiment. The goal is to show what can be created quickly with AI, how temporary and purposeful little apps can be and where this direction of technology is heading. I then write about what I made and share a short demo.

Our first idea in this series is the Amazon Spending Analyzer.

With the run-up to the Christmas season, I, like many others, found myself adding far too many items to my Amazon basket. At the same time, I was dealing with the annual joy of tax returns and wondered whether it was possible to export Amazon purchases and analyse the data.

It turns out you can, although Amazon does not make it obvious. The feature seems to exist thanks to EU data-access rules. If you visit the link below, replacing .com with your local Amazon domain, you can request a full export of your data:

https://www.amazon.com/gp/privacycentral/dsar/preview.html

You can also find it in your account settings under Manage Your Data, where you can request your data directly. Once you go through 8 layers of 2 factor you can request your order history, it takes 1-2 days and you will get an email to download the data.

Once I had the data, I took a quick look in Excel to understand what I was dealing with. I could see product names, dates and times, prices, currencies, shipping costs – everything you would expect from an order history. With that confirmed, it was time to build a dashboard. Using Claude Code, mostly through voice, I described what I wanted. This was the rough prompt:

“A dashboard that summarises what I’ve bought, highlighting things like number of orders, total amount spent, busiest times of day and days of the week. I can see multiple currencies in the file, so include a way to toggle between them if they are present. Also show aggregate data in graph form, particularly around categories of what I bought and the most recurring or most expensive products.”

After a bit of iteration, I had a working report.

I then showed it to a friend who immediately asked how he could generate his own version, so I added another prompt:

“Make it so that a user can select their order history file and, entirely within the browser and without sharing data externally, populate the dashboard.”

That produced the interactive page.

I added a final prompt to generate a sample data file so people could see what they would get, giving them a reason to request their own data from Amazon.

I then deployed it to https://asa.leemallon.com, with a video demo included below. Enjoy.

▶ Watch the video on Substack