CSVLens

Absolutely. Your CSV data never leaves your browser. CSVLens uses DuckDB WASM, which runs entirely in your browser's memory. Your CSV files are loaded directly into your browser's memory, and all SQL queries execute locally on your device. No raw data is transmitted to our servers or any third party. Your data is automatically cleared when you close the browser tab. Only your natural language question (like "Show me revenue by month") and column names are sent to the AI, never the actual data values.

Currently, CSVLens primarily supports CSV (Comma-Separated Values) files. File requirements include a maximum size of 500MB (browser memory dependent), a header row with column names, and standard CSV delimiters (comma, semicolon, tab) which are auto-detected. UTF-8 encoding is recommended. You can upload up to 5 CSV files simultaneously and join them in your analyses. Support for Excel (.xlsx), Parquet, and JSON formats is coming soon.

CSVLens uses advanced AI models to convert your natural language questions into SQL queries. When you ask a question in plain English, CSVLens sends your question along with your table schema (column names and types) to the AI. The AI then generates the appropriate SQL query, which runs locally in your browser against your data. The results are automatically formatted and visualized. For complex questions, you can enable "deep analysis mode" (Agentic Analysis) where the AI breaks down the question into multiple steps and provides comprehensive insights.

Tokens are units of AI processing consumed each time you ask a question for the AI to understand your query and generate SQL. The Free Tier includes 50,000 tokens per month (approximately 150-200 queries), while the Pro Tier offers 1,000,000 tokens per month (approximately 3,000-4,000 queries). Simple queries typically use ~250-500 tokens, complex queries with multiple tables ~500-1,000 tokens, and Agentic Analysis (deep mode) ~2,000-5,000 tokens. Your current usage is always visible in the dashboard, and you'll receive alerts when approaching your limit.

Yes, CSVLens offers multiple export options. For Free Tier users, you can copy results to the clipboard (tables and charts), download query results as CSV, copy generated SQL queries, and screenshot individual charts using browser built-in functions. Pro Tier users gain additional capabilities, including exporting charts as high-resolution PNG images, exporting entire dashboards as PDF reports, batch exporting multiple visualizations, and exporting with custom branding.

You can load and analyze multiple CSV files simultaneously. The Free Tier allows for up to 5 datasets, while the Pro Tier offers unlimited datasets. CSVLens supports cross-dataset analysis, meaning you can join data across multiple tables, run aggregations, and create combined visualizations (e.g., "Join sales_data with customer_data on customer_id and show total revenue by region."). The only practical limit is your browser's memory, with most modern browsers capable of handling hundreds of megabytes of data.

Not at all! The core purpose of CSVLens is to make data analysis accessible without SQL knowledge. You can simply ask questions naturally, such as "What's the average order value?", "Show me monthly revenue for 2024", or "Which customers spent the most?". However, if you do know SQL, you have the option to view and edit the AI-generated SQL queries, which is beneficial for fine-tuning queries, learning SQL by example, or running custom queries directly.

Currently, you can create dashboards by pinning your favorite queries and charts, which can then be exported as PDF reports (available in the Pro tier) for sharing. Team features are actively in development, with upcoming functionalities including team workspaces for collaboration, shareable dashboard links, real-time collaboration on analyses, role-based access control, and SSO for enterprise teams. You can join their waitlist to get early access to these features.