At AI Fund, we’ve watched AI automate more and more of the office work that used to fill people’s days, and out of that we’ve noticed a shift in what actually makes someone valuable, which is that everyone now needs to become more technical.
But “technical” doesn’t mean what it used to. It no longer means learning Python or understanding how a database works. It means developing real fluency with frontier AI, and the strange thing is that the barrier to getting there has never been lower than it is right now.
Across our portfolio companies, the people who are thriving are the ones who understand and actually use frontier AI. They know what LLMs can and can’t do. They understand context windows and hallucinations. And they know the mitigation techniques, how to ground a model’s outputs and how to control error rates through frameworks and human-in-the-loop checkpoints.
The people staying relevant are overwhelmingly software engineers, and it’s worth asking why. I think it’s because engineers learn by building. They read the papers and they go to the conferences, but more than anything they build things, and that’s where the understanding actually comes from.
Building matters here because the conversation around AI is so noisy. Some engineers will tell you AI is months away from recursive self-improvement, and others will tell you LLMs are just pattern matchers that can’t really reason. You can’t sort out who’s right by reading hot takes, because hot takes don’t give you intuition. You only get there by getting your hands dirty and seeing for yourself.
AI is a kind of alien intelligence with jagged capabilities, superhuman along some dimensions and surprisingly weak along others, and developing an intuition for that shape takes practice. You have to sit with context windows and token generation, with where the models shine and where they fall apart. Using AI well is a craft, and like any craft you build it through repetition.
A good place to start is with AI-assisted coding tools. Install Cursor or Claude Code, and pick any project at all, automating a spreadsheet, building a small web app. The specific project matters far less than the fact that you’re building something.
Then make something that actually uses an LLM, a chatbot, a document summarizer, anything. Get a feel for prompts, temperature, token limits. Watch what happens when you send text to a model, and let yourself learn tokenization, context windows, the difference between a stateful and a stateless API. You can skip the math and still grasp the concepts.
When the AI writes code for you, ask it to explain each part, and ask it about the alternatives it didn’t choose. Treat it like a tireless teaching assistant that never gets impatient with you.
This will feel uncomfortable, and you’ll feel slow and clumsy doing it. That feeling is what learning actually feels like.
The barrier to becoming technical has basically collapsed, because AI now handles the programming languages and the syntax that used to take years to absorb. What’s left, the part that’s actually on you, is curiosity and persistence.
And the reach of this is wide. A product manager can prototype features without waiting on engineering resources. A salesperson can automate their own funnel. An operations manager can build their own productivity tools.
Technical fluency is quietly becoming table stakes. The marketing manager who can’t use AI for customer analysis will get outpaced by the one who can. The financial analyst who lives only in Excel will fall behind. The HR leader who doesn’t understand AI screening tools will lose ground without quite knowing why.
The real skill is pairing your own judgment with this alien intelligence, leaning on it where it’s strong and stepping in where it’s weak.
And if you’re just entering the workforce now, you actually have an advantage, because you’re starting fresh in an AI-native world with no pre-AI habits to unlearn. You can build the right intuitions from day one.
So start today. Open Claude or ChatGPT and build something small, a to-do list, a recipe organizer, it doesn’t matter what. Pay attention to where it excels and where it struggles, and let that experience become your intuition.
The young professionals who develop these skills early are the ones who will shape the future of work, because companies need people who understand how to direct alien intelligence. In the end it comes down to time and attention, and while everyone else is still debating whether AI matters, you can quietly be mastering it.