Moving On: Why I Left my job in the Startup (part 1)

The Original Idea

It's been a wild ride, but I've decided to move on from the startup I worked at for many years.

When I first joined, the founders had a compelling vision. They wanted to build a platform to help clients extract financial data from various documents like asset statements and cash flow reports. This data would be aggregated to create a comprehensive financial profile for each high-net-worth client, empowering them to make informed investment decisions. One thing they didn't tell me was that the reason they had this idea was because they were bidding for a project with UBS. About a month after I joined, this idea was scrapped because we didn't get the contract, and the product we built was kinda useless. So, we had to pivot.

First Pivot (consumer wealth tracker):

Since we had the extraction software, which basically spit out a CSV or Excel spreadsheet after the user uploaded a document, and we had the aggregation functionality from the first idea, we thought maybe we could make it a personal wealth tracker. I was a bit against it, as I mentioned many times. I only have two bank accounts, and I can easily track how much money I have. However, my concerns were always shot down by the team, and perhaps the people in the room thought alike, unlike me, the odd one out. I don't understand why people need to track their finances when they only have one or two bank accounts. It made sense for the first idea because high-net-worth individuals have many assets, and it's harder to keep track of everything. But for the average person? I moved along and built this software with open-source Beancount and Fava. We hired a Flutter developer, and I even helped him create the encryption code for encryption in Dart for Flutter. I created the classifier logic, which helped classify each transaction into different categories like:

restaurant -> Starbucks, transport -> Grab

with the help of FastText from Facebook. I also created a tool for non-technical support staff to update the knowledge base for the FastText ML model. In addition, I modified the Fava frontend code to allow users to see their monthly spending and to implement user authentication. With these developments, we applied to the Singapore Fintech Festival and reached the top 40, but unfortunately, we weren't selected for the top 3.

After a few months, there was no traction, and we moved on to our second pivot. There were many reasons why I think it didn't work, as I mentioned earlier, I didn't like this idea. I'll explain it next time...

To be continue