How to Set Up Deepnote for AI Data Analysis

What Deepnote Does and Why It Matters

Deepnote is a collaborative data notebook built for analysts, data scientists, and operations teams who need to explore data without managing local environments. Think of it as a modern replacement for Jupyter notebooks, with built-in AI assistance, real-time collaboration, native SQL support, and one-click data source connections. It runs entirely in the browser, so there is nothing to install.

What sets Deepnote apart in 2026 is its AI-first design. The platform includes an autonomous agent that can write, run, and debug code on your behalf. You describe what you want in plain English, and Deepnote Agent builds the notebook for you, including SQL queries, Python code, charts, and written analysis. For teams that work with data but do not write code daily, this removes the biggest barrier to getting insights out of raw datasets.

Step 1: Create Your Account

Go to deepnote.com and sign up using your Google account, GitHub account, or email. The free plan supports up to three editors, five projects, and basic AI features including ten code completions per month. If you need unlimited AI, the Team plan runs $59 per editor per month and includes a 14-day free trial. Students, teachers, and researchers can apply for the free Education plan, which includes nearly all Team features at no cost.

After signing up, Deepnote drops you into a workspace dashboard. This is your home base for all projects, shared notebooks, and data connections.

Step 2: Create a New Project

Click "New project" from the dashboard. Each project in Deepnote is a self-contained workspace that can hold multiple notebooks, data files, and integrations. Give your project a descriptive name tied to the analysis you plan to run, such as "Q1 Revenue Analysis" or "Customer Churn Investigation."

Inside the project, you will see a blank notebook with a single empty block. Deepnote notebooks use blocks instead of traditional cells. Each block can be a code block (Python or R), a SQL block, a text block, or a chart block. You can mix and match these freely within the same notebook.

Step 3: Connect Your Data

Click the "Integrations" panel on the left sidebar. Deepnote supports over 60 data source connections, including PostgreSQL, MySQL, BigQuery, Snowflake, Redshift, MongoDB, and Amazon S3. Select your data source type, enter your connection credentials, and test the connection.

For quick analyses, you can also upload files directly. Drag a CSV, Excel, or Parquet file into the project file browser on the left side. Uploaded files appear in the project's file system and can be loaded directly into a Python block using pandas:

import pandas as pd
df = pd.read_csv("your_file.csv")
df.head()

If you connected a database, you can immediately start querying it. Create a new SQL block, select your connected integration from the dropdown at the top of the block, and write your query. When you run a SQL block, Deepnote stores the result as a pandas DataFrame that you can reference in any Python block below it.

Step 4: Use Deepnote AI for Code Generation

Every code and SQL block includes an AI button. Click it and type a natural language instruction like "show me monthly revenue trends grouped by region" or "find all customers who churned in the last 90 days." Deepnote AI reads your connected schema, identifies the relevant tables and columns, and generates a working query in the correct SQL dialect for your database.

For Python blocks, the AI can generate data cleaning scripts, statistical analyses, and visualizations. Ask it to "create a correlation heatmap for all numeric columns" or "run a linear regression on sales vs. marketing spend" and it produces executable code using standard libraries like pandas, matplotlib, seaborn, and scikit-learn.

The AI also offers inline code completion powered by Codestral from Mistral. As you type, it suggests the next line or block of code based on your notebook context, including the data you have loaded and the variables you have defined.

Step 5: Use Auto AI for Autonomous Analysis

Auto AI is Deepnote's fully autonomous mode. Instead of generating one block at a time, Auto AI takes a high-level goal and builds an entire analysis workflow. Click the Auto AI button at the top of your notebook, describe your objective (for example, "analyze customer support tickets to find the top complaint categories and their trend over the last six months"), and let the agent work.

Auto AI will devise a plan, then create and execute a sequence of text blocks, code blocks, and SQL blocks. It writes explanatory text between code sections, handles errors by rewriting and re-running blocks, and produces charts and summary tables as part of its output. The result is a complete, readable notebook that you can review, edit, and share.

This feature is particularly useful for exploratory analysis where you do not know exactly what questions to ask yet. Give Auto AI a broad prompt like "explore this dataset and surface the most interesting patterns" and it will run descriptive statistics, check for outliers, test correlations, and generate visualizations on its own.

Step 6: Build Dashboards and Share Results

Once your analysis is complete, you can convert any notebook into a data app or dashboard. Click "Publish" in the top right corner to create a shareable view that non-technical stakeholders can browse without seeing raw code. You can also set up scheduled runs so the notebook refreshes automatically with updated data.

For collaboration, share the project link with your team. Deepnote supports real-time multiplayer editing, similar to Google Docs. Multiple users can work on the same notebook simultaneously, with cursor tracking and inline comments.

Export options include PDF, HTML, and direct push to GitHub or GitLab repositories. If your team uses version control, Deepnote tracks revision history for every notebook, so you can compare changes and roll back to previous versions.

Tips for Getting the Most Out of Deepnote AI

Write specific prompts. Instead of "analyze this data," say "calculate the month-over-month growth rate for each product category and flag any category with a decline greater than 10%." Specificity produces better code on the first try.

Use the AI chat panel for exploratory questions. The chat interface in the bottom right corner is designed for brainstorming and asking follow-up questions about your data without cluttering your notebook with experimental blocks.

Leverage SQL and Python together. Run SQL to pull and filter data from your warehouse, then switch to Python blocks for statistical modeling and visualization. Deepnote makes the handoff seamless by storing SQL results as DataFrames.

If you want to skip the notebook setup entirely and go straight from a file upload to charts and insights in one prompt, tools like VSLZ handle that workflow with no configuration required.

What Deepnote Costs

The free plan is generous enough for individual analysts and small explorations: three editors, five projects, 5 GB RAM machines, and basic AI with ten completions per month. The Team plan at $59 per editor per month unlocks unlimited AI, 16 GB RAM machines, unlimited projects, app publishing, and 30-day revision history. Enterprise pricing is custom and adds features like SSO, custom AI providers, audit logs, and dedicated support.

Summary

Deepnote removes the friction of setting up a local Python environment and connecting to data sources manually. The AI features, especially Auto AI, turn it into a tool where you describe what you want to learn from your data and the platform builds the analysis for you. Create an account, connect your data, and let the agent handle the rest.