Hi, my name is Lukas, and I’ve been a professional Data Engineer for over 10 years. My career has been spent building proper data warehouse setups and replacing the chaos of spreadsheets. Ironically, here I am, writing about how Google Sheets is now my go-to tool for prototyping large language model integrations.
Having built plenty of Python LangChain integrations, I’ve often found the boilerplate annoying. All I want is to prototype some simple, explorative tasks with language models. Sure, ChatGPT’s web interface is decent for quick prompts, but when you need to explore data-driven queries or template chains, it doesn’t quite hit the mark.
Then I realized Google Sheets might be more useful than expected. One day, I sat down and thought, "What if I could just write =Q(prompt) in a sheet?" And there it was: An incredibly simple way to perform OpenAI queries without needing any Python setups. After a bit of fiddling with the Properties Service, for storing the OpenAI key, I had a tool that makes rapid exploration easy.
The more I played with this setup, the more I enjoyed its flexibility. With Google Sheets, I can manipulate data, evaluate outputs, and iterate quickly—all in one place. It’s the best of both worlds: rapid prototyping and data-driven workflows without the usual Python hassles.
If you’re curious, I’ve made the template and the Google Apps Script open source. You can grab it here for free: Google Sheets ChatGPT Function. All you need is an OpenAI API key, and you’re ready to go.
Here is a GIF of Q in action.
With a few simple prompts, I can use the column header to generate content for each column based off the inputs of the adjacent cell.
So here I am, after years of working to move businesses away from spreadsheets, now advocating for them. For this specific use case, at least. Turns out, sometimes the simplest tools really are the best.
Top comments (1)
"the simplest tools really are the best."
Brilliant insight!