In [1]:AI agent for Jupyter

The AI agent that works inside Jupyter.

Ask a question. Runcell writes and runs the code, reads the notebook outputs, and carries the work forward.

Out [1]: from a question to an executed result

live agent replay

runcell — retention analysis
runcell agent connected

ask runcell to work through your notebook

Ask runcell about this notebook…

retention_analysis.ipynb Python 3 · idle

empty notebook — the agent writes it live

In [2]:how it works

From a question to an executed result

Runcell connects the steps that normally fall between prompts: it inspects the notebook, writes and runs the code, reads the outputs, and turns the result into the next useful experiment.

Out [2]: autonomous agent
01Autonomous agent

Turn a question into an executed workflow

Describe the result you need. Runcell plans the notebook steps, writes the Python, executes the cells, and fixes errors as the work develops.

Multi-step workflows, executed for you

Out [3]: in-context assist
02In-context assist

Inspect any cell and unblock the next step

Ask about a transformation, result, or error in context. Runcell reads the surrounding cells and outputs, applies the fix, and keeps the analysis moving.

AI that understands the cells around it

Out [4]: learn by doing
03Learn by doing

Explore methods with runnable experiments

Try analytical approaches side by side with real notebook outputs, then use the evidence to decide which direction fits the question and data.

Concepts explained with runnable cells

In [3]:the agent loop

An agent that works through the notebook with you.

AI code editors can suggest the next function. Runcell works through the sequence that makes a result useful: understand, plan, run, inspect, and continue from what actually happened.

01

Inspect the context

Understand the notebook, data, existing code, and the question before deciding what the next useful step should be.

02

Plan & execute

Break the task into notebook steps, write the Python, run the cells, and recover from errors without leaving Jupyter.

03

Read the outputs

Interpret tables, charts, statistics, and other cell outputs so the next decision is based on what actually happened.

04

Keep moving

Carry the result forward, answer the next question, and turn it into another executed, inspectable notebook step.

In [4]:who runcell is for

Built for notebook-driven work.

For analysts, data scientists, and researchers whose work has to survive beyond a single prompt or a single notebook cell.

field 01

Data analysis

Move from a business or operational question to reproducible code, clear evidence, and a result you can explain.

field 02

Data science

Explore data, test methods, build models when needed, and compare results without hand-assembling every notebook step.

field 03

Risk & quantitative finance

Analyze markets, portfolios, forecasts, and risk with the diagnostics needed to defend the result.

field 04

Research & experimentation

Turn a question into code, figures, and reproducible evidence while keeping the analysis in Jupyter.

In [5]:why runcell

What is Runcell?

Runcell is a Jupyter-native AI agent for data analysis, data science, and research. It turns questions into executed notebook work — inspecting data, writing and debugging Python, running cells, and reading the tables, charts, and other outputs your code produces. Instead of stopping at a suggestion, Runcell carries the analysis forward from the result.

Runs inside JupyterLabWrites & executes PythonBuilds multi-step workflowsReads tables, charts & outputsWorks on existing .ipynb filesNo API key required

what runcell can do

It builds the workflow

Start with a question. Runcell breaks it into notebook steps for inspection, transformation, analysis, validation, and the next useful follow-up.

explore the features or the documentation

It reads the results

Runcell reads tables, statistics, charts, and other cell outputs so it can reason about the evidence instead of guessing from code alone.

It keeps the context

Questions, decisions, and previous outputs stay connected across iterations, so the next step builds on the work instead of restarting from a blank prompt.

Key capabilities

Notebook-nativeRuns inside JupyterLab on your existing .ipynb — no new editor.
Auto execution & debuggingRuns cells, diagnoses errors, and keeps your analysis moving.
Output-awareReasons about the charts and results your cells produce.
Context-awareKeeps questions, outputs, decisions, and next steps connected.

how it compares

AI coding tools vs. a notebook agent

vs. Cursor & AI IDEs

AI IDEs are built around files and source code. Runcell is notebook-native — it executes cells, reads their outputs, and continues from what the code actually produced.

vs. Copilot autocomplete

Autocomplete predicts the next line. Runcell turns a question into a multi-step notebook workflow, runs it, diagnoses the result, and carries the work forward.

vs. notebook AI chat

Notebook chat can explain a cell. Runcell acts on the notebook — editing code, running cells, reading outputs, and continuing the analysis from the evidence.

faq

Frequently asked questions

Runcell is a Jupyter-native AI agent for data analysis, data science, and research. It turns questions into executed notebook work — writing and debugging Python, running cells, reading outputs, and helping you continue from the result.

If you like agentic AI coding but work in Jupyter, yes — but Runcell is designed for notebook workflows rather than file editing. It executes cells, reads tables and plots, and works through multi-step analytical tasks inside the environment where the work runs.

Yes. Runcell keeps cross-session memory of your dataset, your decisions, and the state of your work, so you can pick up a multi-week project days later and ask “what did we do so far?” instead of re-explaining everything.

Yes. Runcell is built for long-running notebook work — from extended analyses to experiments that take many steps — executing them end to end without losing the thread.

Install the extension with “pip install runcell” (or through the JupyterLab extension manager), restart JupyterLab, and sign in. There is no separate desktop app and no new IDE to learn.

No. Runcell includes access to leading AI models such as GPT, Claude, and Gemini based on your plan, so you can start without bringing your own API key.

Yes. Runcell reads the visualizations and image outputs your cells produce, so it reasons about real results instead of guessing from your code alone.

Yes. Runcell works on your existing .ipynb notebooks in JupyterLab and respects your workflow, adding conveniences like a file tree, global search, and git directly in Jupyter.

Yes. The free Hobby plan includes monthly credits so you can try Runcell on a real project, and paid plans add more credits plus access to advanced AI models for heavier notebook work.
runcell demo + intro in 1 minute

Start your next notebook with Runcell

Bring a dataset and a question. Runcell writes and runs the notebook workflow, reads the result, and helps carry the analysis to the next defensible decision.

download the extension

free to start · runs in your JupyterLab · no API key required

Out [6]: product walkthrough

appendix

Notes from the lab

view all posts
tutorialnbconvert tutorial: Convert ipynb to HTML, PDF, and Python scriptsguide5 Ways to Delete a Column in Pandas DataFrameexplainerUnderstanding .ipynb_checkpoints: What they are and how to manage them