After starting to play with ChatGPT seriously in the last few weeks, I'm convinced that any white collar / knowledge worker, who can document all of their relevant background knowledge into accurate technical writing, and learn how to prompt an LLM to read it and return the data in the way their job requires, could automate 80% or more of their work, and actually produce better work, right now, today, with the models we already have. These models are masters at following instructions, and especially good at doing so with large amounts of data, when prompted correctly. It seems like the major limitation right now is people who can prompt them in a way to get out what they want. Incidentally, I just read an article in one of our Aussie newspapers about a guy who got a full refund on a $100k car... by loading all the court cases about consumer car cases into chatGPT, and being his own lawyer. Others said it would have taken years for the guy to have gained the understanding of the law that this guy showed. https://www.google.com/search?client=safari&rls=en&q=how+chatgpt+helping+this+driver+get+a+refund+on+his+ev&ie=UTF-8&oe=UTF-8&safe=active App building is no different! Which makes a company like Lovable raising $150mil at a $2bil val (in progress at least - per TechCrunch) absolutely insane. Who are these investors? Do they know something we don't? Because you've just outlined how to do the same thing using Claude, VSCode, Github etc after 6 months messing around as a small agency. Does Lovable have some magic that us mere mortals cannot comprehend? Or are their investors deluded???