AI-assisted reporting for official statistics
From statistical tables to publication-ready reports.
AtlasStat combines deterministic calculations with agentic AI to produce publication-ready reports. Each release is designed to be reproducible and auditable, with the reporting team retaining final approval.
What is AtlasStat?
AtlasStat is an agentic AI framework for recurring statistical reporting at national statistical offices. It turns validated tables from the standard statistical production process into styled, multilingual publications in Word and PDF.
Prototype · Local testing
The team is testing AtlasStat version 0.2.0 on national accounts reports. Contact us for more information.
How it works?
AtlasStat combines AI agents with deterministic R programs to turn validated production data into a styled report, with translated versions when needed. R calculates the results, and the agents draft and check the text against those results.
Read more about how it works-
Set up the publication
Define the report structure, map source tables and choose where charts and tables go. Reuse this setup for each release.
-
Update the facts and draft
Run the calculations on the latest tables. AI agents draft, check and translate the commentary using the updated facts.
-
Assemble and review
Apply the organization’s style and produce Word and PDF documents for the reporting team to review and approve.
A personal initiative
Who is building
AtlasStat?
I’m Iliass Taleb, a Senior National Accounts Statistician at the High Commission for Planning, Morocco’s national statistical office.
Working on national accounts publications has shown me how much care it takes to update a report each cycle and trace the published figures and commentary back to the source tables. I started AtlasStat to make that workflow easier to repeat and review.
This project is a personal initiative, currently financed by me.
Interested in AtlasStat for your reporting team?
After testing, AtlasStat intends to offer publication setup and collaboration free of charge. Any third-party running costs, such as hosting, model usage or infrastructure, would be discussed beforehand.
Join for availability updates or get in touch about a publication, statistical reporting, R, AI, translation or open-source model evaluation.