10 Feb 2025 - We made our first sale!

We made our first sale!

Then I refunded it, and it cost us $1.31 in credit card processing fees... This is the equivalent of framing [-1.31]. This is a perfect example of failing fast.

As my head engineer,HAMMAD SATTIsays, we must Accelerate Failure to Achieve Fast. He says AFAF in shorthand.

It all happened so quickly. The file was uploaded, and we received confirmation that the process has started - all automation was going to plan. At this moment, I failed and assumed the process would just work. Two 2-hours later - at least 1.5 hours longer than needed - I got an email: "Hi - waiting for my results. When should I get them?".

Everything that could have failed did: 1) the server crashed killing the queue (which we had over-engineered); 2) upon first investigation we were attaching the results to the return email but the file size was so large the SMTP server couldnt attach it (easier fix moving to a d/l link vs attached file; 3) our over-engineered queue was causing memory leakage which caused the entire app to go slow; 4) icing on the cake while the results we shared were correct, there was still false positives and it was hard to find all the issues we suggested.

Yikes. It was quite the end of the week! But we worked over the weekend and came up with a game plan. It won't just resolve the immediate issues above (new queue, improved server-side hardware, link, etc.); it will also improve the core app—better presentation of results, optimization of existing items we check for to improve accuracy, and better early warning notifications.