Looking back, building an AI financial advisor tool for teenagers turned out to be less about the technology and more about the people. When I first set out on this project, I assumed the algorithm would be the hardest part. I was wrong about that. The real challenges showed up in places I had not anticipated: the nuance of language, the ethics of financial advice, and the attention spans of the very audience I was trying to help.

One of the earliest problems was that the chatbot had a serious personality issue. It sounded like an antique tax collector. I quickly realised that no teenager was going to sit through explanations packed with terms like “capital expenditure” when they could just close the tab. If the tool did not speak their language, it did not matter how accurate it was. Nobody benefits from information they never actually read.

Then came what I started calling the regulatory tightrope. In finance, there is a razor-thin line between teaching someone how money works and giving unregulated financial advice. Crossing that line is not a minor mistake. I spent a lot of late nights building guardrails into the system, making sure the AI stayed strictly in the educational lane. The frustrating part was that the safer I made it, the more robotic it sounded. Finding the balance between responsible and genuinely useful turned out to be one of the hardest design problems I faced.

The third challenge was the real rival: the scroll. I was not competing with other educational apps. I was competing with TikTok, Discord, and every piece of content designed to reward attention in fifteen seconds or less. If the bot took more than two or three sentences to get to the point, I had already lost the user. I scrapped the long explanations and replaced them with micro-lessons: short, focused bursts of information built around immediate action and real scenarios. The goal was to make finance feel as fast and reactive as the apps they were already using, not like homework.

This project taught me that the most powerful algorithm in the world is useless if you do not understand the person on the other side of the screen. I started out thinking like a coder. I finished thinking more like a psychologist, a teacher, and a designer at the same time.