How to Get Started with Deepnote for Data Analysis

What Is Deepnote and Who Is It For

Deepnote is a cloud-based analytics notebook that lets you query databases, analyze CSV files, and build shareable reports using SQL, Python, and AI in the same interface. Unlike traditional Jupyter notebooks, Deepnote runs entirely in the browser, requires no local installation, and includes an AI agent that can write and execute code from a plain-English prompt. It is designed for analysts, ops managers, and small teams who need to work with data regularly but do not want to maintain a local Python environment.

According to G2 reviews updated in early 2026, Deepnote is rated 4.4 out of 5 across more than 120 reviews, with users frequently citing the SQL-plus-Python workflow and real-time collaboration as the two strongest differentiators from standard Jupyter notebooks.

Step 1: Create a Free Account and Set Up a Project

Go to deepnote.com and sign up using a Google or GitHub account. Deepnote does not require a credit card to get started. The free tier allows up to 3 editors and 5 projects per workspace, which is sufficient for most individual analysts or small teams evaluating the platform.

Once signed in, click New project in the top-left panel. Give the project a name, such as "Sales Analysis" or "Monthly Report." Deepnote creates a blank notebook inside the project automatically. A project functions as a container: it holds notebooks, connected data sources, files, and scheduled runs together in one place.

Step 2: Connect a Data Source or Upload a File

Deepnote supports two main ways to get data into a notebook: connecting a live database or uploading a file.

To connect a database, open the left sidebar and click Integrations, then Add integration. Deepnote supports over 60 data sources including PostgreSQL, MySQL, BigQuery, Snowflake, Redshift, and MongoDB. Enter your connection credentials and click Test connection. Once connected, the integration appears in every notebook within the project, and Deepnote shows a schema preview so you know exactly which tables are available before writing a single query.

To upload a CSV or Excel file, drag the file directly onto the notebook canvas. Deepnote automatically creates a SQL block with a sample query already written. You can also use the file panel in the sidebar to manage uploads. For tabular files, Deepnote makes the file queryable through its DataFrame SQL engine, which means you can run standard SQL statements against a local file without loading it into any database first.

Step 3: Write Your First SQL Block

SQL blocks are the fastest way to pull data in Deepnote. Click the + button between any two cells and select SQL block. Choose your data source from the dropdown, then write your query.

To use AI to generate the query, press Cmd + K (Mac) or Ctrl + K (Windows) inside the SQL block, then describe what you want in plain English. For example: "Show me total revenue by product category for the last 90 days, sorted descending." Deepnote writes the SQL and explains what it did. You review the query, run it, and the results appear as a table below the block.

The output of every SQL block is automatically stored as a Python DataFrame. This means you can reference the result in a subsequent Python block without any extra code, using the variable name Deepnote assigns automatically.

Step 4: Use the AI Agent for Full Analysis

Deepnote's AI runs in two modes. Generate mode writes code blocks based on your instruction and waits for you to run them. Auto AI is fully autonomous: it writes blocks, runs them, reads the output, and self-corrects until the task is complete.

To trigger Auto AI, open the prompt bar at the bottom of the notebook and type a full request. For example: "Load the CSV I uploaded, calculate the month-over-month growth rate by region, and plot a bar chart." Deepnote adds Python blocks, executes them in sequence, handles any errors, and produces the chart.

This is useful for analysts who know what they want but do not want to look up Pandas syntax. The agent handles standard transformations like pivots, date parsing, and aggregations correctly in most cases. For more complex multi-step analyses, Generate mode gives you more control because you can inspect and edit each block before it runs.

Step 5: Visualize Results with Chart Blocks

Deepnote includes a built-in chart builder that works without writing code. After running a SQL or Python block, click Add visualization below the output table. A no-code chart editor opens. Select the chart type (bar, line, scatter, or pie), map columns to axes, and adjust colors. The chart updates live as you change settings.

For more control, Python visualization libraries work directly inside notebooks: Matplotlib, Plotly, and Altair all render inline. Plotly and Altair charts are interactive by default, letting viewers filter and zoom inside the published notebook.

Step 6: Schedule and Share the Notebook

Once your analysis is complete, you have two options for sharing.

Share a live link: Click Share in the top-right and set the notebook to public or team-only. Recipients can view or, if given edit access, modify the notebook in real time. This is the fastest way to hand off a report.

Schedule automated runs: Open the Schedule panel and set the notebook to run hourly, daily, or weekly. On each scheduled run, Deepnote re-executes every block from top to bottom using fresh data from the connected database. The updated notebook is then available at the same share link. This turns a one-time analysis into a live report that refreshes automatically.

For teams that need clean, formatted output rather than a full notebook view, Deepnote supports exporting to PDF or converting sections to a standalone app view that hides the code blocks entirely.

Common Mistakes to Avoid

One frequent issue is connecting a database at the workspace level rather than the project level. If a database integration is added to the workspace but not enabled inside the specific project, the SQL block dropdown will not show it. Always verify the integration appears in the project sidebar before writing queries.

Another common error is expecting DataFrame SQL to work on files that have not been uploaded to the project file system. If you reference a local file path that exists only on your computer, the query will fail. Upload the file through the Deepnote interface first.

Deepnote vs Jupyter: The Practical Difference

Standard Jupyter notebooks require local installation, manual environment management, and separate tools for sharing results. Deepnote eliminates all three: it runs in the browser, manages Python environments automatically, and includes native sharing and scheduling. For teams where most members are analysts rather than engineers, this removes the setup friction that often blocks Jupyter adoption.

Hex and Marimo offer similar collaborative notebook experiences. Deepnote's clearest advantage is the SQL-first workflow: its native SQL blocks with AI generation are tighter than Hex's SQL cells and more accessible than writing raw Pandas. If your team primarily works with a SQL database and only needs Python for visualization or light transformation, Deepnote is the faster starting point.

If you want to skip the notebook setup entirely, VSLZ handles CSV and database analysis from a file upload with a plain-English prompt, without requiring any SQL or Python configuration.

Summary

Deepnote is a practical entry point for teams that want collaborative, AI-assisted data analysis without managing local Python environments. The free tier is enough to evaluate the full workflow. Connect a database or upload a CSV, generate SQL with the AI prompt, visualize results with chart blocks, and schedule the notebook to refresh automatically. The main investment is the initial data connection setup; once that is done, the turnaround from question to shareable report is typically under 30 minutes for a first-time user.