Ruben Hassid started posting about ChatGPT in late 2022, shortly after it launched. He was working as a ghostwriter, could not find beginner-friendly explanations for people who do not code, and started writing them himself on LinkedIn. That turned into one of the largest AI followings on the platform and then into How to AI, a Substack that now shows over a million subscribers. Search interest in his name has roughly quadrupled over the past year, which is why I wanted to look at it properly.
The format is the reason it works. Each issue is a single walkthrough. He tests a tool on his own machine, takes screenshots, and shows you exactly where to click and what to paste. Typical topics are things like building a NotebookLM notebook from an expert's videos so you can question it like a tutor, setting up a project in Claude, or a specific prompt pattern for a work task.
One recent paid post laid out his method for learning almost any skill fast by finding the best expert content and feeding it to AI. That is genuinely useful for the audience he writes for, which is office workers who are curious about AI but are never going to read API documentation. Most AI newsletters are either news digests or aimed at developers. This sits in a gap that is much bigger than either.
The limits come from the same format. A newsletter is not a course. There is no progression from basics to advanced use, no exercises, no feedback and no sense of having finished anything. You learn whatever this week's topic is.
For someone who wants to properly understand how these tools work, that is a weak way to learn. For someone who wants to pick up one useful habit a week, it is close to ideal. The paid tier is where I have reservations. Some posts are marked paid, including a How to start with AI guide, which seems like exactly the post a free newsletter for beginners should give away.
Substack only shows the price after you enter an email, and I could not find a reliable public figure, so I cannot tell you whether it is good value. My advice is to stay on the free tier for a month or two, see how many paywalled posts you actually wanted, and decide then. The writing also has some of the LinkedIn growth style to it, with bold claims about revenue and followers in places. That is a matter of taste rather than a flaw in the teaching, but it may put some readers off.
My 3.8 reflects excellent free practical content for non-technical people and an unusually clear teaching style, offset by having no structure and a paid offer that is opaque from outside. If you want a structured course instead, Anthropic's AI Fluency course or Google's AI Essentials are better starting points, with How to AI as the weekly top-up alongside them.