The AI-Native Spreadsheet Engine

# Cell B49 — Forecast next 6 months of revenue
import statsmodels.api as sm
import polars as pl

revenue     = A1_A48                       # range → numpy array
seasonality = pl.Series(C1_C48).cast(pl.Categorical)

model = sm.tsa.SARIMAX(
    revenue,
    order          = (2, 1, 1),
    seasonal_order = (1, 1, 1, 12),
    exog           = seasonality,
)

# Output spills into B49:B54 — recomputes only when
# revenue or seasonality changes.
model.fit().forecast(6)

-- Cell A1 — Top regions by Q4 sales
SELECT
    region,
    SUM(amount)                                             AS total,
    AVG(amount)                                             AS avg_order,
    APPROX_DISTINCT(buyer)                                  AS unique_buyers,
    PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY amount)     AS median
FROM   sales
WHERE  quarter = 'Q4'
   AND region IN ('EMEA', 'APAC', 'NA')
GROUP  BY region
ORDER  BY total DESC
LIMIT  20;

You
> Make me a financial model for our Q4 forecast.

Expanse
I'll create a workbook with:

  ✓ Revenue       monthly history + ARIMA projection
  ✓ Costs         fixed + variable, scenario-driven
  ✓ Headcount     by department, monthly
  ✓ Summary       NPV, IRR, runway

Want me to seed it from last year's actuals, or start blank?

You
> Use last year's actuals from finance.xlsx

Expanse
Imported 2024 actuals. Built the model on your existing chart of
accounts and used FORECAST.ETS for the revenue projection.

  ✓ 4 sheets, 1,247 cells, 312 formulas
  ✓ NPV (8% discount):  $4.2M
  ✓ IRR:                27.3%
  ✓ Runway:             18.4 months

The forecast cells are highlighted in blue — click any one to
see the formula and edit it directly.