How you can learn anything in AI | Yacine Learning
TL;DR
AI adoption is mostly a mental-model problem, not a tooling problem — Yacine says enterprises don’t need another generic product as much as they need someone who can map deterministic software, ML, and LLM-style problem solving onto their actual business context, from finance teams to manufacturing plants diagnosing broken vaccine vials.
People anchor hard on their first impression of a technology — he compares AI skepticism to startup pivots at his company AXIA, where old prospects kept thinking “marketplace for suppliers” even after years of evolution into ERP-connected supplier management SaaS, just like users who tried a weak 2024 chatbot and still think that’s what AI is.
The best founders don’t start by “building a product” — they start by helping people — his core startup advice is to do the expensive, manual version first, charge for the value, then turn the repeated work into software, like building $10K websites before creating the $1K website builder that scales the same help.
Feeling like you never fully have the basics is normal, even for experts — Yacine says he’s never met anyone who truly feels they’ve “unlocked” fundamentals, pointing to researchers, lab heads, and even high-end ML people who only know enough at one layer and rely on teams or time to drill down when needed.
His antidote to imposter syndrome is surprisingly literal: go be an imposter on purpose — he tells people to enter domains where they’re obviously underqualified, then learn in public, like when he joined a molecular biology lab without knowing how to pipette, struggled for months, and eventually produced work that made it into a biochemistry paper.
The throughline of his life is not AI or startups — it’s learning itself — from tutoring to a PhD, from reading The Brain That Changes Itself after a push from his swimming coach Christine to teaching his 5-year-old son to read with flashcards built from a favorite book, he frames everything as creating conditions where learning can emerge.
The Breakdown
From AI teacher to enterprise fixer
Yacine opens by describing himself less as a founder than as someone who teaches AI and helps companies actually use it. His current work is an AI consulting shop for enterprise clients, where the hard part isn’t flashy demos — it’s translating AI into useful decisions inside real businesses, especially in manufacturing and finance.
Why most companies still don’t “get” AI
He argues the gap is mostly about mental models: engineers may understand deterministic software, then struggle with ML, and struggle even more with LLMs because the form of problem-solving changed again. His job is often to walk in, see whether a company needs classic software, data-driven ML, LLMs, or a mix, and guide them through that shift.
The AXIA story and why first impressions stick for years
Yacine uses his supply-chain startup AXIA as the big analogy: they started as a supplier marketplace, pivoted repeatedly, and eventually became ERP-connected software for managing existing suppliers. But old prospects kept seeing them as “the supplier-finding company,” which he says is exactly how many people still think AI is just the mediocre chatbot they tried once in 2024.
Why adoption will be slower than AI Twitter thinks
He pushes back on the idea that mainstream adoption will happen overnight. The release cycle feels explosive if you track every model update, but most people don’t — they just remember one bad experience, and he thinks younger users will adopt first while everyone else trails in over time.
Entrepreneurship as scaled help, not product worship
One of the strongest parts of the conversation is his startup philosophy: stop obsessing over “building a product” and focus on helping people. He walks through the logic with a website example — first do the $10K custom work manually, learn what matters, then build the software that lets customers solve the same problem themselves for $1K.
Teaching because he used to be bad at explaining
Yacine says he wasn’t naturally great at communication; he often understood things but couldn’t explain them and came off as dumb until he just built the thing. That gap pushed him into tutoring, conference talks, and eventually YouTube, where he deliberately practiced until he could make ideas legible instead of just having them in his head.
You probably don’t lack the basics — you lack trust in how learning works
When the host talks about being broad but not deep, Yacine basically says: that’s normal. In research, business, and engineering, almost nobody can instantly explain every layer on demand; what matters is being able to move up and down the abstraction stack over time, not having every tensor-level detail cached in your brain at all moments.
His anti-impostor playbook: be shameless, ask dumb questions, keep going
His answer to impostor syndrome is pure Yacine: go do something you’re wildly unqualified for and watch yourself become useful anyway. He tells a story about entering a molecular biology lab with no real wet-lab background, failing for months, barging into offices with broken experiments, and eventually producing publishable work — proof that low ego and relentless learning matter more than belonging.
Learning, children, and the final advice
The last stretch ties everything together: he loves AI because he loves learning, full stop, and traces that obsession from school to neuroscience to ML to startups. He even explains how he taught his older son to read around age 5 by turning a favorite book into flashcards, then closes with his biggest advice: get alone, think hard about what you actually want, find the thread, and use it to decide what to learn and which opportunities to ignore.