Today we are announcing the general availability of our next-generation predictive API, a fully managed time-series forecasting service that has been running in private beta with select customers for the past six months. If your business generates data over time — sales, demand, traffic, sensor readings — this API turns that history into calibrated forecasts with a single call.
The headline change from our previous generation is automatic model selection. Instead of asking you to choose between statistical baselines and neural forecasters, the service benchmarks a portfolio of models against your data's seasonality, trend, and volatility, then serves the winner. Cold-start series with little history fall back to hierarchical priors learned from related series, so even brand-new products get usable forecasts from day one.
Just as important, every forecast now ships with prediction intervals rather than a single point estimate. Planning teams do not need to know that tomorrow's demand will be exactly 1,240 units; they need to know the range they should staff and stock for. Calibrated uncertainty turns a forecast from a guess into a decision-making tool.
During the beta, customers used the API to cut inventory holding costs, plan staffing for seasonal peaks, and flag anomalous demand shifts days earlier than their previous tooling. The service is available now, with transparent usage-based pricing and no infrastructure to manage. We can't wait to see what you build with it — and if you have a forecasting problem that doesn't fit the mold, talk to us.
Back to all articles