wzn1118/AsteriaAnalyst

What it solves

Asteria Analyst is a local workbench for corporate data analysis, statistical modeling, and the delivery of formal management reports. It bridges the gap between raw business data (like sales, customer, or inventory tables) and professional, traceable reports, ensuring that AI-generated insights are backed by deterministic calculations and verifiable evidence.

How it works

The system uses a structured pipeline to move from raw data to a final PDF report. It employs a "mandatory AI chain" that handles semantic mapping of fields, business context routing, and metric planning, while leaving the actual numerical calculations to a deterministic executor to ensure accuracy.

Key components include:

• Analysis Lab: A workspace with thousands of "method cards" for statistical experiments, allowing users to bind fields and run specific tests (e.g., t-tests, regression, or clustering).
• Revision Center: A workspace for reviewing, annotating, and versioning reports before final publication.
• Local Extensions: Support for custom "Skills," agent teams, and a Codex Runtime for reusable analysis workflows.

Who it’s for

It is designed for business analysts, data scientists, and management consultants who need to transform operational spreadsheets into formal, audited management reports with a high degree of traceability.

Highlights

• Deterministic Execution: Ensures numerical accuracy by separating AI planning from actual computation.
• Extensive Statistical Library: Includes over 360 registered statistical methods and thousands of method cards for various analysis types (descriptive, diagnostic, predictive, etc.).
• Traceable Workflow: A strict pipeline from data profiling to PDF release, including evidence validation and semantic mapping.
• Local-First Approach: Runs locally on Windows, supporting local AI providers and data privacy.
• Integrated Revision System: Dedicated tools for tracking changes, adding annotations, and managing report versions.