How to Get Started with Hex for Data Analysis

Hex lets you connect a database or upload a spreadsheet, describe the analysis you need, and get a finished, interactive report — without switching between a SQL editor, a notebook, and a dashboard tool. The Notebook Agent handles query generation and chart creation from natural language. Setup takes under 15 minutes.

What Hex Does and Who It Is For

Hex is a browser-based analytics workspace that lets data teams and analysts work with SQL, Python, and no-code tools in a single notebook environment. Unlike traditional BI tools that separate data transformation from visualization, Hex keeps all of it in one place. The result is a unified workflow: connect to a database or upload a file, write or generate the analysis, and publish a shareable app or report directly from the same interface.

Hex is useful for analysts who work with a data warehouse and want to move faster without switching between a SQL editor, a Jupyter notebook, and a dashboard tool. For non-technical users, the Notebook Agent provides a natural language layer that handles query generation and chart creation automatically.

According to G2 reviews in 2026, teams that adopt Hex report cutting report turnaround time by roughly 40 to 60 percent compared to manual SQL-plus-dashboard workflows, largely because stakeholders can interact with finished apps directly rather than requesting one-off queries from an analyst.

Setting Up Your Hex Account

Hex is entirely web-based. There is nothing to install. Go to hex.tech and sign up with a work email or connect via Google or GitHub SSO.

The free Starter plan gives you one workspace user, unlimited projects, and access to the Notebook Agent with a limited number of AI credits per month. The Team plan at $75 per user per month adds collaborative editing, scheduled runs, and higher AI usage limits.

After signing in:

1. Name your workspace. Use your company name or team name; this appears in shared links.
2. Choose a data source. Hex prompts you to connect a database or upload a file in the onboarding flow. You can skip this and connect later.
3. Explore the demo project. Hex pre-loads a sample dataset and notebook. Run through it to see how cells, charts, and apps interact before working with your own data.

Connecting Your Data

Hex supports two main data input methods: direct connections to cloud databases and file uploads.

Cloud database connections. Hex integrates natively with Snowflake, BigQuery, Redshift, Databricks, PostgreSQL, and MySQL. To connect:

1. Go to Settings, then Data Connections, then Add new connection.
2. Select your database type.
3. Enter your credentials: host, database name, username, and password (or OAuth for Snowflake and BigQuery).
4. Click Test Connection to confirm the setup works.

Once connected, Hex indexes your schema and makes table and column names available to the Notebook Agent as context, which improves the accuracy of AI-generated queries.

File uploads. For one-off analysis, you can upload CSV, Excel, or Parquet files directly into a project. Click the data icon in the left sidebar of any project, then Upload a file. Hex loads the file into an in-memory DuckDB table that you can query with SQL or reference in Python.

File uploads work for most spreadsheet analysis tasks but are not suitable for files larger than a few hundred megabytes. For large datasets, connect directly to a warehouse.

Creating Your First Project

A project in Hex is a notebook that contains your analysis and the app that gets published. Each project starts with an empty notebook. You add cells in different types: SQL, Python, markdown, charts, input parameters, pivot tables, and more.

To start:

1. Click New project from the dashboard.
2. In the notebook, add a SQL cell by clicking the plus icon and selecting SQL.
3. Choose your data connection from the dropdown in the cell.
4. Write a query or click Ask AI to describe what you want in plain English.

The Notebook Agent interprets your description, writes the SQL, and runs it. For example: "Show me total revenue by month for the last 12 months" generates a working query from your schema with no additional configuration.

Using the Notebook Agent for AI-Assisted Analysis

The Notebook Agent is Hex's core AI layer, running on Claude Sonnet 4. It can generate and run SQL queries, write and execute Python code, create charts with specific styling, build interactive parameters, and restructure your notebook into logical sections.

To use it, click the wand icon at the top of your notebook and describe the analysis you want:

"Analyze this sales data and show me top 10 customers by total spend with a bar chart"

The agent will add cells, generate the code, and run them. If the output is wrong, describe the correction in the same chat thread without starting over.

A few prompting approaches that produce better results:

• Be specific about the output format. "A bar chart sorted descending by revenue" produces a more accurate result than "visualize this."
• Name the columns. If your table has a column called order_total_usd, use that exact name in your prompt rather than referring to "the revenue column."
• Use the graph-aware context feature. If you have multiple cells with related queries, tell the agent "using the data from the previous SQL cell" to chain steps together correctly.
• Iterate instead of redoing. If a chart is mostly right but missing a label or filter, ask the agent to fix only that part rather than regenerating the entire analysis.

A December 2025 update to the Notebook Agent added support for chart styling, input parameter creation, and automatic cell cleanup. These allow the agent to not just write queries but also organize and polish the final notebook, reducing the manual work of formatting reports before sharing.

Building and Sharing an App

Every Hex project has two modes: the Notebook (where you build and edit) and the App (what viewers see). The App is a clean, interactive interface that shows charts, tables, and input parameters, hiding all the code.

To build the app view:

1. Switch to App layout using the toggle at the top of the project.
2. Drag and drop the elements you want viewers to see: charts, table cells, metric callouts, and input parameters.
3. Input parameters are particularly useful. A date range picker, a dropdown for customer segment, or a text search box can all be wired to SQL queries so the report updates dynamically.

When the app is ready, click Publish and share the link. Viewers can filter and explore the data without editing any code. If you want to require login, set the sharing permissions to workspace members only.

Scheduling Reports

For reports that need to run on a regular cadence, Hex supports scheduled runs on Team plans. Go to Run settings in the project toolbar and set the frequency: hourly, daily, weekly, or a custom cron expression.

A common use case is a weekly operational report that runs Sunday night, populates with the latest data from the warehouse, and notifies stakeholders via email that a new version is ready.

Limitations Worth Knowing

Hex's pricing puts it in the same range as Tableau Creator at $75 per user per month on the Team plan. For teams that need collaborative SQL notebooks plus reporting and AI in one tool, this is reasonable. For solo analysts who primarily want dashboards, a lighter BI tool may be a better fit.

The Notebook Agent has a monthly credit limit on most plans. Complex multi-step analyses that call the agent repeatedly can exhaust credits quickly. For large notebooks, it is worth planning AI calls in advance to avoid interruptions mid-analysis.

If your primary workflow involves uploading CSV or spreadsheet files and analyzing them with natural language questions rather than connecting a warehouse, VSLZ AI handles the same task without any setup: upload a file, describe the analysis, and get charts and summaries from a single prompt.

Practical Summary

Hex works best for teams that already use a cloud data warehouse and want a single tool for analysis, code, and reporting. The setup takes under 15 minutes: create an account, connect your database or upload a file, and start a project. The Notebook Agent handles SQL and chart generation from plain-English prompts. When the analysis is ready, the App layer turns it into a shareable, interactive report that non-technical stakeholders can filter and use on their own.