VerbaGPT
VerbaGPT is an artificial intelligence tool designed to allow users to interrogate their data using natural language. The solution is built...
Last verified:
What is VerbaGPT?
VerbaGPT is an experimental tool that enables users to analyze tabular data using natural language queries powered by large language models, while ensuring complete data privacy. Users can upload CSV or TXT files, connect to SQL databases like MSSQL, MySQL, or PostgreSQL, and ask questions in plain English to generate insights, plots, and models without writing code. It leverages Text-to-Python capabilities, going beyond simple SQL to include statistical analysis and visualizations, all running locally on the user's hardware so no data rows are sent to external LLMs.
Key features include advanced analytics for complex queries and data modeling, intuitive design with upload and connect options, curated LLM models including free ones, offline mode (experimental), SQL command blocking for safety, data description management, retry and analyze buttons for error handling and summaries, support for multiple LLM providers like OpenRouter and OpenAI, training with successful query-code pairs, and flagging failed generations.
It is designed for non-technical users seeking preliminary insights from their data and technical professionals aiming to boost productivity. VerbaGPT bridges the gap between data needs and coding skills, helping organizations become more data-driven without compromising privacy or relying on data teams for every query.
VerbaGPT pricing
Pricing model: Free
Trial: Free full-featured 60-day evaluation. Personal: Free after trial, limited to datasets under 10,000 rows, personal use only. Commercial: Contact for unlimited use, priority support, commercial licensing.
VerbaGPT pros
- Runs entirely locally for data privacy
- Supports CSV and TXT file uploads
- Connects to MSSQL, MySQL, PostgreSQL
- Text-to-Python for advanced analytics
- Generates plots and visualizations
- Performs data modeling tasks
- Curated list of free LLM models
- Experimental fully offline mode
- Blocks harmful SQL commands
- Manage descriptions for databases
- Retry button for code errors
- Analyze button for result summaries
- Toggle between LLM providers
- Train model with query-code pairs
- Flag unsuccessful generations
- Simple intuitive interface
- Free for personal use
VerbaGPT cons
- Experimental offline mode unreliable
- Limited to small datasets in personal plan
- Personal plan under 10,000 rows only
- No row data access by LLM
- Requires schema sharing
- Primarily tabular data focused
- Needs user-approved schema details
- Commercial requires contact
- No PostgreSQL in early mentions
Frequently asked questions about VerbaGPT
How does VerbaGPT ensure data privacy?
VerbaGPT runs locally on your hardware, sending no data rows to the LLM—only explicitly approved schema parts are shared, keeping sensitive information private.
What data sources does VerbaGPT support?
It supports uploading CSV or TXT files and connecting to MSSQL, MySQL, or PostgreSQL databases locally or remotely.
Is VerbaGPT just Text-to-SQL?
No, it's Text-to-Python, enabling SQL queries plus statistical analysis, visualizations, plots, and data modeling.
What are the pricing plans?
Free 60-day full-featured trial, then free Personal plan for small datasets under 10k rows (personal use), Commercial plan via contact for unlimited features and support.
Can I use VerbaGPT offline?
Yes, there's an experimental fully offline mode, plus local hardware execution without sending data rows externally.
Who is VerbaGPT designed for?
Non-technical users for preliminary insights and technical users for productivity gains in data exploration without coding.
What safety features are included?
It blocks harmful SQL commands and allows flagging unsuccessful generations for better control.
How do I handle code errors?
Use the Retry button to regenerate code and the Analyze button for AI commentary on results and code.
Can I train the model?
Yes, train with successful question-SQL or Python pairs to improve future generations.
What LLMs can I use?
Curated list including free models, toggle between providers like OpenRouter, OpenAI, with data description management.