Somewhere between your last team meeting and now, Google slipped something genuinely interesting into Google Sheets. A forecasting AI — one trained on 100 billion real-world data points — is now sitting inside the spreadsheet tool most of us already use every day.
It's called TimesFM, and the idea is simple: you give it your past data (sales by month, orders by week, whatever you track), and it makes an educated prediction about where things are heading. No specialist required. You click through a settings panel, and the model does the rest.
What's notable is that this AI doesn't need to be trained on your data before it can help you. It already learned patterns from an enormous slice of the real world, so it can look at numbers it's never seen before and still give you a sensible forecast. That's rare, and it used to require a data team to pull off.
Forecasting — predicting demand, anticipating slow seasons, planning stock — has always been something only bigger companies could do well, because it required people who knew how to build models. This puts a working version of that capability into a free Google account.
It won't replace good judgment. But it might help you walk into a conversation with your supplier, your bank, or your team with something more than a gut feeling.
Time-series data — any numbers that are tracked over time. Monthly revenue, weekly visitors, daily orders. If it has a date attached, it's probably time-series data.
Zero-shot forecasting — the AI makes predictions on data it's never seen before, without any setup or training. Like asking a seasoned chef to taste your dish and guess the recipe — they just know.
Foundation model — a large AI that was trained once on a huge amount of data and can then be applied to many different tasks without starting from scratch each time.
If you already use Google Sheets to track anything over time, it might be worth pulling up that data and seeing what the model says. Not to follow it blindly — but to see if it notices something you haven't.