What product managers actually need to know about LLMs (and what to do with them)

AI isn't magic, it's math. Building an AI feature is not about falling behind if you haven't done so yet, but rather about understanding where you sit on the AI map. AI is a broad umbrella with various branches, including machine learning and natural language processing. Instead of building towards buzzwords, product managers should work backward from the problem they're trying to solve. If users are dropping out mid-way through setup or your product tour is not converting well, AI isn't the strategy, but rather one tool in the box that needs a well-scoped job. The "Buy-Borrow-Build" framework can help product managers decide whether to buy, borrow, or build an LLM. Borrowing APIs like OpenAI, Claude, or Perplexity is often the smartest approach unless proprietary data and urgent problems require building from scratch. Ultimately, it's the quality of your data that matters more than the model itself. Product managers should focus on framing the problem, scoping the dataset, and keeping engineers honest rather than trying to build models.