The evolution of the data notebook

Hex is building the successor to the traditional data notebook. While Jupyter and its descendants redefined how data scientists write code, they often created silos where insights remained trapped in static, linear documents. Hex addresses this by turning the notebook into a collaborative, reactive platform that allows users to build and share interactive data applications. Founded in late 2019 and based in San Francisco, the company has attracted significant attention from major venture firms like Andreessen Horowitz and Sequoia Capital for its ability to unify the analytical workflow.

Reactive logic and app building

The core of the platform is a reactive execution engine. In a traditional notebook, code runs in a top-down sequence; if a user changes a variable in an early cell, they must manually re-run every subsequent cell to update the state. Hex treats code more like a spreadsheet. Changes propagate automatically through a visual logic map, which provides a Directed Acyclic Graph (DAG) view of how data flows through the project. This architectural choice ensures that the output is always consistent with the code and allows for the creation of interactive elements—like sliders and input boxes—that non-technical stakeholders can use to explore data themselves. This shifts the notebook from a private scratchpad to a public tool for an organization.

The shift to agentic analytics

Since 2023, Hex has pivoted its roadmap toward what it calls "agentic analytics." This shift is embodied in "Magic," its suite of AI-powered tools, and its more recent "Notebook Agent" initiatives. While many competitors offer a sidecar chat window for generating code snippets, Hex is focused on a more integrated experience. The agent is designed to understand the full context of a project, including the database schema, the existing code logic, and the user’s ultimate goal. It doesn't just suggest a line of SQL; it can build out entire sections of an analysis or refactor complex workflows based on high-level intent. This focus involves building a scalable agent infrastructure that allows these models to interact with data warehouses safely and accurately.

Market position and integration

The platform occupies a distinct space between the raw flexibility of a Python script and the rigid constraints of business intelligence tools like Tableau or Looker. It is designed for analysts, scientists, and engineers who need to move quickly from a raw query to a polished, published app. For companies like Mercor, which used the platform to identify significant revenue opportunities, the value lies in this speed-to-insight. Hex connects to the modern data stack, including Snowflake, BigQuery, and Databricks, acting as the primary interface where both humans and AI agents collaborate to make sense of information. By managing the search and context architecture necessary for LLMs to reason about data, Hex is positioning itself as a central hub for the next generation of automated data work.