Sphinx AI is a copilot for anyone working with data. Sphinx AI Copilot currently ships as a VSCode extension that interfaces with Jupyter and other compatible notebooks. Our copilot runs in your environments, alongside you. All compute invoked by Sphinx AI runs on your servers and is subject to your permissions.
Sphinx AI builds data intuition by running inference jointly across modalities ranging from natural language, to images, to statistical representations. By carefully choosing the right representations for your data and the task at hand, Sphinx AI delivers highly contextualized outputs accelerate your workflow.
Sphinx AI is always learning from you; the more you use it, the better it gets. Sphinx AI can also be configured by natural-language rules (both globally and local to each of your projects)
Sphinx AI can access anything you can reach through Python, REST APIs, or MCP including Snowflake, Databricks, BigQuery, Salesforce, and more. By running locally in your environment, Sphinx AI generates artifacts and insights that you can audit, reproduce, and fully own.
Agent Units are a measure of compute directly proportional to the amount of inference Sphinx AI needs to use to complete your tasks. A single Agent Unit is enough to complete a simple agentic task, such as building and validating a robust linear model. You can monitor your usage of Agent Units on the Sphinx dashboard.