[Organizational AI Utilization #208] It is recommended to use spreadsheets as a database in a Google Workspace environment

Hello! This is Terada.

Currently, I serve as the Representative Director of AI Digital Community (ADC), a community for AI practitioners such as AI promoters at digital-related companies, as well as the Representative Director of FURIKAKE Partners Inc., which supports the AI-transformation of client businesses, and AI Portalize Inc., which provides products that support organizational AI utilization. I support organizational AI utilization from various perspectives!

Recently, I have the impression that the number of companies using AI to automate internal workflows or develop simple web tools in-house is increasing rapidly.
There is an active movement where frontline staff are taking action themselves to streamline daily routine tasks.

However, I often see cases where they hit a wall regarding where to store data or how to link multiple tools.
I believe that building a full-scale database from scratch is a high hurdle both technically and in terms of cost.

Especially for companies that have introduced a Google Workspace environment, this challenge can be solved smoothly by applying tools that are already familiar.
Therefore, this time, I would like to write about using spreadsheets as a database when automating workflows with AI, while touching on concrete examples.

Example 1. Data aggregation in the creation of minutes for business negotiations and meetings

When conducting business negotiations via Google Meet or similar, recording data and raw transcripts are saved to Google Drive.
A flow can be considered where transcription and summarization are performed using Gemini, utilizing tools like "Google Workspace Studio" mentioned in previous articles.
In that case, it is very effective to aggregate the output results into a spreadsheet once, rather than sending them directly to a chat tool.

※Please also refer to this article for specific methods of automating AI workflows by combining Google Workspace Studio and Gems.

For example, organize the data with the meeting title in the first column, the date in the second, the full transcript in the third, and the summary in the fourth.

The image is to list the scattered information from each meeting as a single tabular database.
By keeping the original information in a spreadsheet, it becomes easier to reuse that data later.

In other words, it becomes easy to perform integrations such as extracting only the next actions from the accumulated data into a separate column, or sending the results to specific people via email.

If you want to perform additional processing by citing the main text every time, it is very effective to use this as a simple database.

Of course, I think the point is not to save all data aimlessly, but to create columns narrowed down to items you want to use for search and analysis later.

Furthermore, by listing the data, I believe that information sharing across the entire team will also become smoother.

This is because if you only send it to a chat tool, past information will flow away, making it difficult to look back.
As a result, I have the impression that having information accumulated in a fixed location called a spreadsheet also leads to the resolution of reliance on specific individuals.

Example 2: Data storage and hub function in web app development

This is the second example.

When creating internal web apps using Google Apps Script, data storage and accumulation are also necessary.

In such cases, a configuration where all information is put into a spreadsheet and read/written to is recommended.
Essentially, you can have the spreadsheet play the role of a database.

This is because spreadsheets have excellent compatibility with various Google services.
Even without preparing a dedicated server, you can safely perform permission settings and access management with standard Google Workspace functions.

In addition, it becomes easy to link with external tools such as Slack via Google Apps Script.

※Please also see this article regarding how to deploy internal AI tools using GAS and Google Sites, as well as security settings.

For example, suppose you create a simple help desk app to automate responses to internal inquiries.

The situation would be that the user's question content is recorded in a spreadsheet, the AI creates a draft answer and writes it in another column, and that is notified to Slack.
It is ideal to use a spreadsheet as a "hub" to aggregate information, and then turn it into an app or incorporate it into a workflow from there.

※Please refer to this article for details on the mechanism to automatically cultivate internal FAQs with AI and automate even primary responses.

Conversely, I believe it is safer to start small by using a spreadsheet as a base before embarking on full-scale system development.
It is easy for on-site staff to check the data content, and it is also easy to add or modify items while in operation.

By consolidating data into a spreadsheet, both the speed of development and the scope of subsequent utilization will expand significantly.

Operation and maintenance of automated tasks and integration into business processes

I have been explaining the benefits of setting up a spreadsheet as an intermediate point, rather than simply passing information output from AI directly from one place to another.

In short, by structuring and accumulating data, it becomes much easier to automate continuous tasks.

Also, from the perspective of operation and maintenance, inserting a spreadsheet in the middle has great significance.
There are cases where AI processing stops midway or the expected output is not obtained.

Even if an error occurs, if the original data remains in the spreadsheet, you can see at a glance where the processing failed.
You can smoothly re-execute it as is, and it can also be used as test data to improve AI prompts.

I think it would be best to start by trying to consolidate minutes from familiar meetings or building simple notification workflows.

Keeping data in a form that humans can see and understand, rather than just as a temporary transit point, is the key to increasing the accuracy of AI utilization.
Using these as a starting point, please try to find data accumulation methods and automation mechanisms that suit your company.

Thank you for reading!
Here are past AI-related articles!

By the way, I have also published a book summarizing AI promotion in organizations. Please take a look if you like.