As controversial as it may sound, half the people who talk and write about artificial intelligence has never worked on any serious (or just toy) project. They talk AI (and machine and deep learning) by plain example, whether they are lawyers, sales, digital experts of some smoky kind, mid- and C-level executives. The other half of people who talk and write about AI are low-level developers, who can split in four the hair of a Python algorithm but hardly see the solution in the context of some real-world business. The purpose of this talk is combining together a long-term vision (both towards the past and the future) of technology with rock-solid knowledge of the software dynamics to identify some milestones of development. AI is simply software to write that doesn't come out of a pre-packaged product. From the customer's perspective AI is just software; from the development perspective it is a in-house platform to develop having clear problems, domains, and achievable results. It's (primarily) a matter of algorithms and it's not a matter of hype.