Claude became a research partner, not just Google 3.0

Before AIM26, I filled my retirement with the discipline I'd spent a career building, but for a while that discipline didn't have a project to attach itself to. I retired in 2015 after nearly four decades as a chemist, finishing a PhD at 49 that I'd started and shelved decades earlier. The early retirement years were, honestly, harder than I expected. What brought the momentum back was online learning — starting with YouTube, to eventually find Mindvalley and some other courses, and then, this year, AIM26. I enrolled because I wanted a real use case, not a demo. For years I'd been quietly collecting multi-language (French, Dutch, English, German) herbs & plants and phytotherapy newsletters with a long-term goal of turning them into a proper, evidence-graded reference and eventually a self-published herbal-health book. I hoped AIM26 would teach me enough to make some AI helpful, how I did not know then, besides some playing around with ChatGPT. I learned about Claude, and his Cowork and Code things And lo and behold, Claude is now more as an actual research partner on that project rather than some Google 3.0. AI Mastery 2026 did more than that. Lesson by lesson, the course gave me a working structure — memory, a knowledge base, reusable skills, quality rules — and I applied some of them to my own material instead of a toy exercise. By the time I reached the end and this coming Awards, what had started as three digitized newsletters had grown into a real relational database. The biggest change, however, has not been technical. It is how I now think about AI as a collaborator. I am a chemist, and somewhat nerdish and “a pietje precies” like they said in Dutch, “a precision obsessed Peter”. I don't hand over judgment easily, and I was genuinely skeptical an AI could be trusted with primary sources without cutting corners. I watched this AI holding the line. For example, excluding promotional material instead of padding the numbers, refusing to fabricate a newsletter "in the style of" a real author just to hit scale faster, flagging its own data-quality mistakes before I found them myself (ouch for my ego). That is not what I expected from software! It is why I'd recommend this Mindvalley Quest to anyone who wants AI to build something real and ending with a genuine AI partner. My top three takeaways: AI is most valuable not as an answer machine but as a disciplined executor of rules you set yourself; real progress comes from applying each lesson to your own material immediately or asap, not filing it away; and intellectual honesty scales — the same "real sources only, no shortcuts" principle that guided my PhD thirty years ago turned out to be exactly the rule an AI agent needed too. My honesty works both ways and orders me to admit that the pace of the webinars is fast. To my embarrassment it is too fast for me actually. I am behind in assimilating the lessons material. Thankfully, materials remain available online for the next two years. Since finishing this stretch of the course, I feel something I hadn't expected to feel again in my mid-seventies: genuinely ahead of a deadline thanks to Claude AI, on a project that matters to me, with a stack of real evidence to show for it. As are some lessons, the book is still ahead of me — the full newsletter archive is still waiting on a NAS connection I haven't solved yet — but that feels like a logistics problem, not a wall.