A One-Click Installation on your Mac to Use A.I. to Analyze your own Excel File --- No Cloud, Totally Private

A few weeks ago, I wrote about an experiment — running AI locally on your own notebook to analyse spreadsheet data. The response from readers was clear: this is useful, but can you make it easier to use?

So I did.

Today I am releasing the Spreadsheet AI Analyzer as a proper Mac application. Not a Python script. Not a terminal command. A real .dmg file — drag it to your Applications folder, double-click, and start asking questions about your data in plain English.

Why This Matters: The Cloud Problem Nobody Talks About

Every "AI for Excel" tool I've seen has the same silent assumption baked in: your data goes to someone's server first.

For most consumer use cases, that's fine. But if you work in finance, legal, or any professional setting where your spreadsheets contain client data, revenue projections, or privileged information — that assumption is a non-starter.

I built this specifically to remove that assumption entirely.

Your data never leaves your machine. Full stop.

No OpenAI account. No Anthropic API key. No cloud subscription. The AI runs locally, powered by Ollama — the same engine I described in my earlier posts.

What It Does

The concept is simple: upload an Excel or CSV file, and have a conversation with it.

"Which product line had the highest growth in Q3?" "Are there any missing values in this dataset?" "Summarise the trends across all sheets."

The AI reads your data, reasons about it, and streams the answer back to you — word by word, just like the big cloud services, except the entire process happens on your own hardware.

It supports .xlsx, .xls, .xlsm, and .csv. Multi-sheet workbooks work too.

Choose Your Own Brain

One of the things I enjoy most about this approach is that you are not locked into any single AI model. The app ships with a first-run wizard that lets you pick from three:

• qwen3:1.7b (~1.1 GB) — Fast, lightweight, works on any Mac. My recommendation for most people.

• llama3.2:3b (~2.0 GB) — Meta's model. Excellent at plain-English summaries and explanations.

• qwen3:8b (~5.2 GB) — The heavyweight. For complex financial modelling and deep pattern recognition. Needs 8 GB+ RAM.

You can add more models later directly inside the app — no terminal required.

Zero Technical Setup

I was very deliberate about this. The earlier experiment required Python, a terminal, and some patience. This version requires none of that.

Install it like any other Mac app. The AI engine, the backend server, everything — starts automatically when you launch the app and stops cleanly when you close it. No processes left running in the background.

First launch note: macOS may show an "unidentified developer" warning since the app is not yet code-signed. Right-click → Open to proceed. This is standard for independent software at this stage.

Try It

👉 excelaianalyzer.netlify.app

The landing page has the download links for both Apple Silicon and Intel Macs. The app is free. Questions and feedback welcome at ordinaryaijedi@gmail.com.

The tools are free. The privacy is total. The potential is limited only by your imagination.

Future Enhancement

Let me know if there is any desired features you want. Will be happy to explore. In the meantime, enjoy.