You probably have at least one spreadsheet that you open, stare at for a few minutes, and then close without really knowing what to do next. Maybe it’s a bank export with too many transaction categories, a product list with missing prices, or a survey file where some answers are blank. Numbers and Excel can help, but when the file is large or the columns are messy, You end up clicking through filters and formulas instead of getting the answer you actually need. On a Mac, this can feel like a bigger chore than it should be.
What most people want is a way to ask a plain question about the file — like “which rows are incomplete?” or “what is the total per category?” — and see the result without building a complicated spreadsheet model. Online tools can do this, but uploading private data to a website is not always acceptable. A local option is to run a Jupyter notebook on your Mac, which keeps the raw data on your machine. A tool such as RunCell — a Jupyter-native AI agent — can turn your written request into working code, execute it, and even interpret the results, so you don’t need to learn Python first. This article shows a practical, privacy-friendly way to check a spreadsheet and get a repeatable result.
Start With One Clear Question
Before opening any app, write down the one thing you want to know from the file. Instead of “understand this export,” choose a decision like “find all orders that are missing a customer email” or “see which products have no price entered.” This question will guide everything else and prevent you from wasting time on columns that don’t matter.
Then open the file in Finder and take a quick look at its raw structure. If it’s a CSV file, right-click it and open it with TextEdit to see the first few lines and confirm that commas separate the fields. If it’s an Excel file, open it in Numbers just long enough to identify the column names you need. Write down the exact names of the columns that relate to your question. This small step makes the later steps much easier because you won’t be guessing at labels.
For very large files, consider exporting a smaller version from the original app if you can. A bank statement with five years of transactions will slow down any tool, so export only the month or quarter you need. The goal is to keep the file small enough that your Mac handles it quickly.
Set Up a Private Jupyter Notebook on Your Mac
Jupyter is a free tool that runs in your web browser but works with files on your Mac. It uses “notebooks,” which are documents that combine text, code, and results in one place. You don’t need to be a programmer to use a notebook for a simple data check, especially when an AI helper like RunCell writes and runs most of the code for you.
1. Install JupyterLab, pandas, and RunCell in an Isolated Folder
First, create a new folder for your spreadsheet project and put the export file inside it. Open Terminal on your Mac and type cd followed by a space, then drag the folder into the Terminal window and press Return. Now run these commands:
# Bash
python3 -m venv review-env
source review-env/bin/activate
pip install jupyterlab pandas runcellWhen the installation finishes, launch JupyterLab with:
# Bash
jupyter labYour browser will open a new tab with the Jupyter interface. In the file list on the left, click “New” and choose “Python 3” to create a blank notebook.
Note: RunCell appears as a side panel or chat window inside JupyterLab. If you don’t see it, check the left sidebar or the extensions menu.
2. Load the Spreadsheet and Keep It Small
In the first cell of the notebook, paste this simple code and run it by pressing Shift+Return:
# Python
import pandas as pdNow you have the pandas library ready. The next cell will load your file, but you should only load the columns you need. For example, if your question is about missing emails and order totals, and your file has columns named customer_email and order_total, run:
# Python
df = pd.read_csv('yourfile.csv', usecols=['customer_email', 'order_total'])If your file is an Excel workbook, replace read_csv with read_excel and include the sheet name if needed. This keeps the notebook fast and focused.
3. Ask a Plain-English Question with RunCell
Here’s where an AI agent saves you from writing code. Instead of typing Python, open the RunCell panel and describe what you want in plain English. For example:
“Show me all rows where customer_email is missing or order_total is zero, and return the row number and the reason.”
RunCell will generate the pandas code, run it in your notebook, and display the resulting table — all without you touching the code. Because it’s a Jupyter AI agent, it can also read the output and adjust if something looks wrong. Before you trust the result, compare the number of returned rows with what you expect from a quick filter in Numbers, and check that the column names in the generated code match your file. This verification step matters because AI can misunderstand a request, especially when column names are ambiguous.
Turn the Check Into a Repeatable Routine
Once the notebook gives you a useful answer, save it with a name that describes the check, such as missing-email-review.ipynb. The next time you receive a similar export, you can copy the notebook, open it, change the file name in the second cell, and run all cells from the top. In less than a minute, you’ll have the same check applied to the new file.
To make the routine even more helpful, add a small text cell at the bottom of the notebook where you write what you found and what you did. A short note like “Three rows missing email; moved them to a follow-up folder” is enough. This creates a simple log that you can read next time, so you don’t have to remember the context from scratch.
If the export format changes, don’t rewrite the entire notebook. Adjust only the cell that loads the file or the prompt that asks the question. For example, if a column name changes from customer_email to email_address, update that one word in the load cell and rerun the notebook. Keeping the notebook small means you can find and fix these changes quickly.
Keep Your Data Local and Your Workflow Simple
The main advantage of this setup is that your raw spreadsheet files stay on your Mac. You are not uploading customer records, financial data, or personal spreadsheets to an online service. The Jupyter notebook runs locally in your browser, and the file itself never leaves your machine unless you choose to share it.
One important nuance: when you use the AI features in RunCell, the tool sends your question and a small amount of relevant context (like column names or a few rows) to the RunCell cloud so it can call the language model. The actual data file is not uploaded, but the prompt and selected context are transmitted. If you’re on a paid plan, RunCell’s Privacy Mode ensures your inputs are not used for training or improvement. If you’re on the free plan, your inputs may be used to improve the service. For most personal analysis this is acceptable, but if you’re working with highly sensitive data, you may want to use a paid plan or avoid AI features and write the code yourself.
You don’t need to become a Python expert to benefit from this approach. Start with one simple check, run it on a single file, and keep the notebook as a template. Over time, you can add more checks or ask slightly more detailed questions. The habit of writing down one clear question before opening the file will help more than any advanced technique. When the notebook gets confusing, return to that question and simplify. A small, local, repeatable check is often all you need to turn a messy spreadsheet into a clear answer.
