Most "AI integration" is a wrapper around one API call.

Most of what I use AI for never reaches a user. It's the research pass before a design starts, a rough first draft of copy that I then rewrite, or a script that turns forty rows of client data into forty personalized assets so I'm not doing it by hand at midnight. For my own business, I built a pipeline that pulled and scored more than 600 institutional leads across four states.

Sometimes a product has to summarize something, classify it or answer a question in plain language. I design those features the way I design anything else. I work out what happens when the model is slow, wrong, or confidently making things up before launch, so a user doesn't find out first. More on this in my article AI Is Not Autopilot.

Wiring an LLM into a product is the easy part. What takes time is a prompt structure that holds up beyond the three examples you tested, a cost and latency budget, a fallback for when a call fails, and usually OpenRouter rather than a single vendor, so the product isn't tied to one company's pricing.

This is where AI has changed my own work the most. Faster first passes in Figma and Stitch, code scaffolding that I review line by line afterward, and Apify pulling reference material I used to collect one tab at a time. It hasn't replaced any of the decisions. There are just fewer hours between having an idea and having something on screen to react to.