Goodbye, App

First, let's talk about book selection. My current approach is to open four AIs simultaneously—GPT, Gemini, Claude, and the expert mode of Doubao. I tell them my focus areas, keywords, and industry, and let them recommend the most recent and valuable books worldwide. Each AI provides a list of books, I take their intersection, then look at the complement. Honestly, the list filtered this way is more accurate than any book blogger’s recommendations. Book bloggers recommend 20 books a month, and they might have only read summaries themselves. AI can read 20,000 books in a second and directly generate a ranking. I no longer worry about not having good books to read; I simply can't finish them all.
Next is how to read. Feed the books to AI, and let it extract the most essential 5,000 words. Over a day, easily cover more than 10 books. For most books of 100k words, the truly useful part for you won't exceed 1,000 words. AI can precisely dig out the part you need most right now. Turning the act of reading ten thousand books from a slogan into an actionable process.
But none of this is the main point.
The key is, if you can do this with reading, it means almost everything can be done this way.
If an app can't allow an agent to access it, in the AI world, it’s as if it doesn't exist. AI can't see it. Users are no longer humans; they are agents. Agents don't care how beautiful your splash screen animation is, and they don't...