Similarly, I had done socket work and reversing proprietary protocols, and I didn't consider any of that groundbreaking (nor was it substantially marketable). To walk into a field like AI and summarization which isn't infantile and outdo Ph.D. level experts at their own theory -- that's something you didn't do, nor did Nick, nor does anyone. It just doesn't happen, and hasn't in any subject with a similar depth of specialization as far as I'm aware.
Not to say one can't make useful things and/or learn a lot very rapidly. I tend to understand the young, inexperienced people building games that sell well due to the frameworks available. And I understand how shareware could sell decently in a market that's widely open with lots of low-hanging fruit. But this situation is very different. That he built an application that grabbed parts of text and produced a shortened version by making some very rudimentary guesses about what's important perhaps. What's described -- not a chance.
Yes, with 3 years of programming experience your code won't be too polished/elegant but it will be sufficient to make the application work. (Especially if you can use some 3rd party frameworks like the kid in this article possibly did.)
That included learning how the POP3 and SMTP protocols worked to write email clients and obviously socket programming.