How to Get Started with Hex for Data Analysis

Hex is a browser-based analytics workspace where SQL, Python, and AI work together in one notebook. You sign up at hex.tech, connect your data source or upload a file, and start querying within minutes. The Notebook Agent writes and fixes code from plain-English prompts, so analysts who are comfortable with data but less comfortable with code can still run complex analysis. Teams can also publish results as shareable interactive apps without any extra engineering work.

What Hex Is

Hex describes itself as "the AI Analytics Platform for your whole team." In practice, that means a notebook environment similar to Jupyter, but hosted in the browser, with live collaboration built in and AI that understands analytics context.

A Hex project has two views: the Notebook view, where you build your analysis cell by cell in SQL, Python, or R, and the App view, where you arrange outputs like charts, tables, and filters into a polished interface you can share with stakeholders. Both views live in the same file. There is no separate export step.

Hex connects to most major data warehouses, including Snowflake, BigQuery, Redshift, and Databricks. It also accepts direct file uploads in CSV, Excel, and Parquet formats, as well as integrations with cloud storage providers like S3 and GCS. If your team stores data in a Google Sheet, Hex can read that too.

Step 1: Sign Up and Set Up Your Workspace

Go to hex.tech and click "Get started." Hex offers a free Community plan and paid plans starting at the Team tier. The Notebook Agent is available to Editors and above on all paid plans, currently in public beta.

Once inside, you land in your workspace home. The left sidebar shows your organization's projects, any collections you belong to, and a data browser that lists all connected data sources. The settings pane is accessible from the bottom left; this is where admins add team members, configure SSO, and manage billing.

Hex provides several demo datasets built into the workspace, so you can explore the interface without connecting real data during evaluation.

Step 2: Connect Your Data

Click "Data" in the left sidebar, then "Add a connection." Hex shows a list of supported integrations. For a direct database connection, provide the host, port, credentials, and a connection name. Hex tests the connection before saving it and encrypts credentials at rest.

To start with a file instead, open any project and click "Upload file" in the data browser panel. Hex accepts CSV, Excel (.xlsx), and Parquet files up to 1 GB. Once uploaded, the file appears as a named dataset available throughout that project. You can reference it in SQL cells with SELECT * FROM your_file_name without writing any import code.

For teams using dbt, Hex integrates with the dbt Semantic Layer and can pull in documented models directly, so analysts work with the same semantic definitions the data team already maintains rather than raw table names.

Step 3: Create Your First Project

From the workspace home, click "New project." Hex opens a blank notebook with one empty SQL cell and a connection selector at the top. Choose your data connection, then write a SQL query or ask the Notebook Agent to draft one.

A basic starting query to profile a new dataset looks like this:

SELECT
  column_name,
  COUNT(*) AS row_count,
  COUNT(DISTINCT column_name) AS unique_values
FROM your_table
GROUP BY 1
ORDER BY 2 DESC;

Run the cell with Shift+Enter or the play button. Results appear inline as a scrollable table. Click any column header to sort, filter, or chart the data directly.

To add a Python cell, click the "+" button below any existing cell and select Python. Python cells have access to the output of any SQL cell as a named dataframe. If your SQL cell is called "sales_query," in Python you write df = sales_query and the dataframe is ready to work with. Pandas, NumPy, Matplotlib, and Plotly are available without any installation.

Step 4: Use the Notebook Agent

The Notebook Agent is Hex's AI assistant that runs inside your notebook. As of early 2026, it uses Anthropic's Claude Sonnet 4.5 and is rolling out to all paid customers.

To open the agent, press Cmd+K on Mac or Ctrl+K on Windows in any cell, or click the AI button in the top toolbar. A prompt bar appears. Type what you want in plain English, for example: "Write a query that shows monthly revenue by product category with a rolling 3-month average compared to the same month last year."

The Notebook Agent reads your connected schema automatically, so it knows your table names and column types before it writes a line of code. It builds the query, explains what it did, and places the result in a new cell. You can ask it to revise, add a chart, fix an error, or explain a line of code without leaving the notebook.

One published use case from Hex's blog: a pricing team used the Notebook Agent to run a multi-table margin analysis that previously required a senior analyst to structure. The agent drafted the full query, including window functions for rolling comparisons, in a single prompt. That translated to a task that previously took several hours being completed in under 20 minutes.

The agent also works on Python cells. Describe a data transformation in plain English, and it writes the Pandas or NumPy code. If that code throws an error, you can ask the agent to debug it in the same session.

Notably, the Notebook Agent does not hallucinate column or table names. Because it reads your actual schema, it works only with columns that exist. This makes it more reliable for production analytics than general-purpose AI coding assistants that lack schema context.

Step 5: Build and Share a Data App

Once your notebook contains the charts and tables you want to share, switch to the App view by clicking the "App" tab at the top. Hex converts your notebook outputs into a drag-and-drop layout canvas.

Drag a chart into position, resize it, add a dropdown filter linked to a variable in your SQL query, and add a date picker to control the date range displayed. No front-end code required. The result is an interactive dashboard that stakeholders can open in a browser, apply filters, and see results pulled from your live data connection.

To share, click "Publish app." Hex generates a shareable URL. Viewers granted access can open the app and apply filters. Scheduled runs let the app refresh automatically on a daily or weekly cadence and notify viewers by email when a new version is ready.

Hex also supports embedding apps into external tools such as Notion pages, internal portals, or Slack via the Hex Agent in Slack integration, which lets users query data in natural language directly from a Slack channel.

What Hex Does Not Do Well

Hex is not a replacement for full BI platforms like Tableau or Looker if your organization needs pixel-perfect reports, complex row-level security at scale, or enterprise governance features. The free plan limits execution compute and does not include the Notebook Agent.

Hex also assumes at least basic familiarity with data concepts. While the Notebook Agent lowers the barrier substantially, users with no exposure to SQL or dataframes will face a steeper learning curve than with tools that operate entirely through point-and-click menus.

For teams that want to skip notebook structure entirely and analyze a file through a plain-English conversation, VSLZ handles uploads end-to-end from a single prompt with no cells or schemas to manage.

Pricing as of 2026

Hex offers a Community plan (free, limited compute), a Team plan, and Enterprise pricing. The Notebook Agent is included in Team and above. A 14-day trial is available for all paid tiers. Exact pricing is listed at hex.tech/pricing.

Summary

Hex is a strong choice for data teams that want a collaborative notebook with AI assistance built in. Setup takes under 15 minutes: sign up, connect a data source or upload a file, create a project, and use the Notebook Agent to write your first query. From there, build interactive apps and share them across your organization without additional tooling.