Five books for the age of agents and not only for a long time I wrote book reviews. After the interview I was asked to recommend five books. I thought about it and recommended. almost five. There were no LLMs, and three out of five were released in 1997. Everything is still relevant, maybe even more than before. 1 Cal Newport, Deep Work (2016) and Slow Productivity (2024) In an era of multitasking, when multiple agents are working at the same time, and you jump between them, these books are more relevant than ever. The first is deep concentration: high quality results require long blocks of uninterrupted work, otherwise superficial solutions are obtained. The second is how to achieve great results over a long distance without burning out: do less at the same time, work at a natural pace and be obsessed with quality. Approximately the long distance is similar to the known coding: 1.01365 ≈ 37.8 and 0.99365 ≈ 0.03. Both books raise the question of how to measure the efficiency of intellectual work. The usual exchange, when overt work is taken away as a result, Newport calls pseudo-productivity. The mission is still not solved, and with the arrival of agents she will have to rethink. Jack Copeland (ed.), The Essential Turing (2004) A collection of key Turing works with Picker's notes and important historical context. Turing is known to almost everyone as the author of the famous test of the ability of machines to think. Many have heard of the Enigma hacker, in which he made a great contribution (see also the post about Polish scientists and the beautiful Poznań Museum). Almost every IT expert knows that Turing 90 years ago proposed a model of computing, now known as the Turing machine. No less people remember that in the same work he proved the existence of unattainable tasks, and in the article with the test there is a section on learning machines, where he proposes to make a "children's machine" and teach it with rewards and punishments, that is, in fact describes RL.