User guide
Kontext Labs is the AI-native workspace for data analysis. Connect your data, bring your own LLM provider, and go from data questions to SQL, charts, and automated workflows — all in one workspace.
This guide is task-oriented: each page covers one workflow, with the screens and controls you need. Start with installation if you are new, or jump to the page for the task you are doing.
Get started
Section titled “Get started”- Install and start Kontext Labs — Install the
kontext-labspackage, start the local server, and replace the placeholder profile. - Open or create a project — Projects hold data sources, files, chats, and settings.
- Connect a data source — Connect an external database or inspect the internal lakehouse.
- Browse and manage files — Upload, organize, search, and open project files.
Notebooks
Section titled “Notebooks”- Work with notebooks — Use text, table, SQL, code, chart, diagram, and image blocks in one document.
- Chart query results — Every chart type, with the data and configuration saved in the document.
- Python and data frames — Run Python blocks and load SQL results with kontext.load_data, without re-querying.
Agents and models
Section titled “Agents and models”- Chat with an agent — Start a conversation and revisit previous sessions.
- Ground an agent in knowledge — Curate the business concepts agents cite, and review their proposals.
- Connect a model provider — Add a local or cloud model and choose the default.
Workspace settings
Section titled “Workspace settings”- Account — Review and edit the local profile that owns this workspace.
- Python environment — Inspect the runtime that executes notebook and chat Python blocks, and prepare it.
- Project settings — Edit the project name and description, and understand local access.
- Jobs and run details — Monitor background runs and inspect their parameters, logs, and timing.