Government statistics
Complex samples, custom tables and scheduled production runs, on machines you control, with an audit trail for every AI request and one policy file for every install.
Methods
Design-correct estimates and standard errors with weights, strata and clusters, and design effects.
Publication tables with nesting, stacking, significance letters and multiple response sets.
Case weights across procedures and tables.
Time series models, including VAR, for indicators and projections.
Rules that find out-of-range, missing and inconsistent values before anything is published.
Every table rebuilt from syntax, run by hand or on a schedule.
Every module is in every plan, with 148 procedures in all. See how the engine is validated.
Controls
Run syntax jobs from the command line on a schedule with launchd, cron or the Windows scheduler.
One policy file sets AI, cloud and endpoint rules for every install, and always wins over user settings.
What the AI saw and what was approved, exportable for governance.
Works offline for up to 30 days between licence checks; permanent offline activation is coming.
A typical project
SPSS, Stata, SAS, CSV and Parquet files with labels and missing codes intact.
Set weights, strata and clusters once for design-correct estimates.
Custom tables with the statistics and tests your publication needs.
Save the syntax and run it as a headless job for each release.
Every module and the private AI, free for 14 days. Institutions can plan a pilot with us.