Everything a researcher needs — upload data, chat with AI, build visualizations, and run repeatable analysis recipes.
Register an account — Create your account at /register. You'll need to verify your email address before logging in. During registration you can set your preferred language and other preferences.
Configure connections — Head to Settings → Database Connections to save your PostgreSQL or MySQL credentials. Then go to Settings → AI Connections to point the Workbench to an AI provider (local Ollama or a remote OpenAI-compatible service).
Upload or connect data — Upload CSV/Excel files under Tools → Upload CSV/Excel, or connect directly to database tables under Tools → Connect Database.
Start analysing — Chat with the AI assistant, build visualizations, or create a repeatable recipe.
Ask questions, get coding help, or brainstorm analysis approaches. Supports audio speech input and output. Manage multiple conversations and switch between them. Copy responses with one click.
Connect a database and let the AI inspect its schema — tables, columns, foreign keys, and indexes. The AI will flag structural issues (missing indexes, data type concerns, normalization problems) and suggest improvements.
Select a dataset and the AI will review column schemas — types, null counts, unique counts, and sample values — then suggest cleaning steps (handling missing values, fixing data types, normalizing text).
Tell the AI what you want to learn from your dataset and it will propose analysis rules — statistical tests, aggregations, groupings — that you can incorporate into a recipe.
Drag-and-drop CSV or Excel files. Configure delimiter, quote character, and encoding. Preview data before importing. The Workbench stores uploaded data in DuckDB for fast local analysis.
Browse tables and views from your configured PostgreSQL or MySQL connections. Select specific columns to pull in. Define them as datasets for analysis. Your database credentials are stored per-user and never shared.
Search and filter all your datasets by source type (upload, table, view, virtual) and group visibility. View metadata, browse data snapshots, and manage column definitions. Trigger data preparation rules from here.
Upload .ipynb files. The Workbench extracts DataFrame definitions, shows column info and row counts, and imports the resulting tables into your database. Great for bringing in analysis work from external tools.
Create interactive Plotly charts — bar, scatter, line, pie, histogram, and more. Choose a visualization type, provide data as JSON or CSV, configure parameters, and render. Export charts as images for reports.
Build repeatable analysis workflows. A recipe bundles data preparation rules and analysis steps. Select a dataset, configure input columns, set filter conditions, add custom parameters, then reorder steps via drag-and-drop. Run the recipe now or save it for later — and get the same results every time on updated data.
Track every execution of your recipes. See when they ran, on which dataset, and review the results. Compare runs over time as your data changes.
As a normal user you cannot: