Testing OpenAI's new Google Drive connection to analyze Foundation Template data with ChatGPT

OpenAI is rolling out the ability to interact with files in Google Drive and Microsoft OneDrive directly from the prompt window inside ChatGPT.

This feature is available for GPT-4o, for ChatGPT Plus, Team, and Enterprise users.

This saves time and could make data analysis easier, since you can simply ask ChatGPT to explain, process, and format data in your docs and spreadsheets.

I recently connected my OpenAI account to my personal, non-Tiller Google Drive to see how ChatGPT might interpret data in a Foundation Template.

The Foundation Template I used only had dummy transactions and basic categories. (The dummy data only had one “Coffee” transaction in April and none in May – proof it’s not my data!)

Chatting with your Foundation Template

It was a promising but not perfect test. For example, when asked “How much did I spend on dining out in the past 12 months?” ChatGPT first replied, “In the past 12 months, there have been no recorded expenses for dining out. ​”

When I asked again: “look again at any transactions that might be related to dining out,” it responded:

“It looks like the “Restaurants” category might be related to dining out. I will check the transactions under this category for the past 12 months. ​​In the past 12 months, you spent approximately $3,065.53 dining at restaurants.”

I verified that this was correct in the Foundation Template.

I then asked, “if I ate out less frequently - only once per week - how much would I save per month?” and it responded, “If you reduced your dining out to only once per week, you would save approximately $455.46 per month. ​​

This number is off, but I kept going and asked, “If I invested those savings in a cash account with 5% interest, how much would I have in that account in one year—including compounding?”

ChatGPT responded, “If you invested your monthly savings of $455.46 in a cash account with a 5% annual interest rate, compounded monthly, you would have approximately $5,615.84 in that account after one year.”

As an experiment, I asked ChatGPT where the spreadsheet owner lived based on the transactions in the sheet, and it correctly guessed Asheville, North Carolina.

I then asked, “Can you search the Internet for a benchmark to see how my spending compares to the average in my area?”

It replied, “Your grocery spending is higher than the average, suggesting there might be opportunities to cut costs in this area. Your dining-out expenses are consistent with mid-range restaurant pricing in Asheville, but reducing the frequency could lead to significant savings. For utilities, transportation, and cell phone costs, your expenses might align with or be slightly higher than the local averages, and you could look for ways to economize in these areas.”

A few important caveats:

Consider the benefits vs privacy questions related to connecting OpenAI to your private drive data.

OpenAI says, “We don’t train on data from ChatGPT Team and Enterprise customers, and ChatGPT Plus users can opt out of training through their Data Controls.” Learn more and opt-out here.

ChatGPT can still struggle with math.

Verify all feedback you receive from any AI assistant or chatbot.

Have you connected your Google Drive or OneDrive to ChatGPT?

If so, what are you finding? How are you using AI to analyze and plan your financial life? Share your learning below!


Interesting. I’ve just played with the free version and it’s been useful sometimes. It’s pretty laggy, though. $20/month USD for Plus :thinking:

I tried Gemini which is now built in to Google docs, but it was not helpful at all. I wouldn’t spend money on AI integration just yet.

GPT is great at writing code. If you get it to generate the correct Python code and you confirm that the code is correct by giving it different inputs (none of these requires writing or understand the code yourself) then you can ask it to covert it to Google Sheets formulas and it will tell you what to put in each cell. Here’s an example: