Loading…
The same statsmodels + sklearn stack a data-science team would write — but driven by a one-line prompt.
Forecast / Sales 90d
Outlook + anomalies
30d forecast
+14%
±6%
Anomalies
2
last 30d
MAPE
4.8%
model fit
Months or years of orders, sessions, costs. There's signal in there.
'Are we going to hit the number?' 'Is anything weird going on?'
Linear extrapolations in Excel and gut-feel calls. Both wrong.
Any table with a date column and a metric — orders.created_at + amount, sessions.day + count, you name it.
'Forecast revenue for the next 6 months' or 'detect anomalies in last quarter's transactions'.
Statsmodels + sklearn fit the model. No notebooks to set up, no Python to install.
Forecast lines with confidence bands; anomalies highlighted on the timeline. Pin to a report.
ARIMA / SARIMA / Holt-Winters / linear-trend models picked automatically based on data shape.
Statistical (z-score, IQR) + ML-based outlier detection. Anomalies are highlighted on the chart.
Forecasts come with upper/lower bounds so you don't mistake noise for signal.
Models detect daily/weekly/monthly cycles and project them forward.
Forecasts and anomaly views drop into reports like any other widget.
Refine the forecast in chat — 'redo it with only the last 12 months' — without re-coding.
Forecasts cash and revenue for board prep without spinning up a data-science team.
Spots anomalies in funnel metrics before the weekly review.
Predicts staffing demand from historical traffic patterns.
Free tier covers most teams. Connect a database in 30 seconds.
No credit card required · 1M tokens / month free