Every product roadmap now has "add AI" on it, and most of those features will be quietly removed within a year. Not because the technology doesn't work — it works remarkably well — but because it was bolted on to look modern rather than to solve a real problem. The teams that win with AI aren't the ones with the fanciest models; they're the ones who picked the right job for a model to do and engineered around its weaknesses. Integrating AI well is less about models and more about product judgment. Here's how we approach it.
Start with a job, not a model
The failure pattern is always the same: a team decides to "use an LLM" and then hunts for somewhere to put it. That's building a solution and looking for a problem, and it reliably produces features nobody keeps. Reverse it. Look at where your users waste time, get stuck, abandon a flow, or stare at a blank field, and ask whether a language model genuinely helps there.
Good candidates share a recognizable shape:
Unstructured input — free text, documents, messy data a rule can't parse.
