At a recent InfoQ presentation, Meryem Arik laid out what she thinks software engineering looks like in 2030: managing what companies spend on AI usage, running many AI agents at once, non-technical people building their own tools, and engineers shifting from writing code toward product decisions. It is a prediction piece, not a news event.
So let us be honest about what that means for a business in Cave Creek with a website and a booking form. Most of it means nothing this year. Predictions about 2030 are a way for an industry to talk to itself about where the work is going. They are not a warning you need to act on.
One thread is worth your attention, though, because it has already arrived. Arik talks about token spend management. A token is roughly a chunk of text an AI tool reads or writes, and most AI services bill by the token. In plain terms: AI features are metered, like water. If someone adds an AI chatbot or an AI-written product description tool to your site, the cost is not a flat monthly fee. It goes up when usage goes up. A quiet month costs little. A busy month, or a bot hammering your chat widget, costs more.
That is the practical difference between an AI feature and, say, a contact form. The contact form costs the same whether you get two enquiries or two hundred. An AI feature does not. If nobody has explained that to you before switching one on, you can get a surprise.
The other prediction, that non-technical people will build more of their own tools, is already half true and already has a familiar catch. Building something is the easy part. Someone still has to maintain it, keep it secure, and know what to do when it breaks at 6pm on a Friday.
What to do: nothing today. But if your site has an AI chat widget, an AI search box, or AI-generated content running automatically, ask whoever set it up two questions: what does this cost per month at current traffic, and is there a spending cap on the account. If the answer to the second is no, ask for one.
Reported by:
