VerbaGPT

VerbaGPT is an experimental data analytics tool that enables natural language querying of your CSV, SQL data, or personal notes, with an emphasis on privacy. It goes beyond basic data aggregation by enabling plotting, complex queries, and data modeling directly from natural language, effectively functioning as Text-to-Python in addition to Text-to-SQL. The system is designed to run locally, ensuring your data never leaves your hardware. Only the database schema (table and column names) and any user-authorized column details are shared with the LLM, preserving data confidentiality.

How VerbaGPT Works

• You interact in plain English to ask questions or request analyses of your data.
• The tool translates requests into Python code and SQL queries, executes them locally, and returns results.
• It supports basic data operations (queries, aggregations), as well as advanced capabilities like plotting and statistical modeling.
• The architecture ensures the LLM does not have direct access to your data; it only receives structural metadata you approve.

How to Use VerbaGPT

1. Provide your data source (CSV, SQL database, or notes) locally.
2. Ask a question in natural language (e.g., “What is the average sales by region this quarter?”).
3. Review the generated results, charts, or models. Refine by asking follow-up questions.

Privacy and Security

• Data never leaves your device; the LLM only receives schema details and user-approved column metadata.
• Designed for local execution to protect strategic assets and sensitive information.

Core Capabilities

• Local, privacy-preserving data analysis
• Natural language queries translated to Python/SQL
• Basic data aggregation and filtering
• Plotting and visualization generation
• Data modeling and advanced analytics (e.g., regression, classification)
• Text-to-Python + Text-to-SQL hybrid workflows
• Schema-aware interaction: only shared metadata is exposed to the LLM
• Experimental software aimed at expanding what’s possible with SME data analysis

Safety and Licensing Notes

• Experimental software; use for exploration and personal analysis.
• Ensure your data handling complies with relevant privacy and security policies.

Example Use Cases

• Summarize key statistics from a dataset and produce a visualization.
• Build simple predictive models from tabular data.
• Execute multi-step queries that combine filtering, grouping, and aggregation.
• Analyze notes or documents by extracting structured insights from unstructured text.

Copyright 2023 – VerbaGPT