Featured On :
AI is bad at data. This startup can fix that  Read now

The Premier AI Agent for Data

Get Started

Fluent in Data

Sphinx understands data, not just code, and uses representation learning to always build appropriate models and algorithms.

Automate LLM-powered Analysis

Sphinx can use native batch inference and embedding functions for sentiment analysis, metadata extraction, and more.

Connected to Everything

Sphinx is easily configured with natural language to connect to essentially any data source. It’s like MCP or Agent Skills, but easier and far more powerful.

An Agent with Superpowers

Kernel context

Sphinx is natively designed to work with Jupyter notebooks, and is given extensive context on defined variables, dataframes, and imports.

Guardrails

Configure guardrails for Sphinx to require user approval before running certain cells.

Hang detection

Sphinx can detect when a cell has been running for too long, and will interrupt and optimize before trying again.

Web Search

Sphinx automatically searches the internet for public datasets and APIs that it can use in performing analyses.

PDF Parsing

Sphinx learns with you, taking notes from past analyses to apply to future projects.

Get Started in Minutes

Sphinx works directly in your existing environments

Sphinx Extension

The Sphinx Extension can be installed into VSCode, Cursor, Windsurf, and more for a best-in-class IDE experience with Jupyter notebooks

Sphinx Dashboard

New: Sphinx Lab

The easiest way to get started, Sphinx Lab provides a ready-to-go Jupyter environment and agent directly in your browser

Sphinx Lab

Sphinx CLI

For power-users and programmatic invocations, run Sphinx directly in your terminal analyses to streamline future work

Sphinx CLI