# VerbaGPT

> VerbaGPT connects frontier LLMs to your databases and files, turning natural language into Python-powered analysis, charts, and statistical models — without writing SQL or code.

## What VerbaGPT Does

VerbaGPT is a natural language data analysis platform. Users ask questions in plain English and receive charts, tables, statistical analyses, and machine learning models generated automatically from their databases and files.

It is text-to-Python (not just text-to-SQL), meaning it can run pandas, scikit-learn, matplotlib, and Plotly code — not just database queries.

## Two Modes

**Virgo** — Cloud-powered analytics via the web app at app.verbagpt.com. Connects to PostgreSQL, MySQL, MSSQL, and Snowflake databases. Also supports uploaded files: CSV, Excel, JSON, PDF, DOCX, Google Drive imports. Uses frontier LLMs via OpenRouter, Groq, and Cerebras. Designed for teams and B2B SaaS deployments.

**Taurus** — Local AI agent powered by the Claude SDK. Runs on the user's machine with full file system access. Designed for data scientists doing complex, multi-step exploratory analysis where data never leaves the local environment.

## Key Features

- **Data Notes**: An automated data dictionary layer. Users describe their tables, columns, and business logic in plain English. This context is injected into every LLM prompt, dramatically improving query accuracy.
- **Prompt Libraries**: Org-wide collections of proven, effective prompts. Teams share expertise; consultancies package domain knowledge as licensed prompt packs for clients.
- **Smart Schema Selection**: Embedding-based schema filtering — only relevant columns are passed to the LLM, reducing token cost and improving accuracy.
- **Multi-model support**: Virgo can use different LLMs for different tasks (schema reasoning, code generation, review). No vendor lock-in.
- **Security**: JWT authentication, role-based access control (user/admin/owner), AES-256 encryption at rest, TLS 1.3 in transit, audit logging.
- **Organizations**: Multi-tenant. Teams share data notes, prompt libraries, and footnotes within their org.

## Who It's For

- Business analysts and managers who need data insights without writing SQL
- Data teams who want to give non-technical colleagues self-serve analytics
- Consultancies that want to productize domain expertise as prompt libraries
- Data scientists who want agentic, local AI workflows (Taurus)
- Enterprise teams who need security, governance, and multi-user support

## Supported Data Sources

Databases: PostgreSQL, MySQL, Microsoft SQL Server (MSSQL), Snowflake
Files: CSV, Excel (.xlsx/.xls), JSON, PDF, DOCX, PPTX, TXT, Markdown, Google Drive

## Company

VerbaGPT LLC. Web app: https://app.verbagpt.com. Landing page: https://verbagpt.com. Contact: team@verbagpt.com. LinkedIn: https://www.linkedin.com/company/verbagpt-llc. Twitter/X: https://x.com/VerbaGPT.

Member of Microsoft for Startups.

## Pricing

Free tier available with no credit card required. Pro and Team plans available inside the app at app.verbagpt.com.