When it comes to AI Agents, Gemini is slowly but surely taking the lead over OpenAI, at least for my workflows. And it isn’t necessarily because Google’s models are better than OpenAI’s or Anthropic’s; rather, it’s due to the flexibility of Gemini AI Agent implementation for those using Google Workspace. I, like many other small business owners, rely on Google Workspace over Microsoft to house my internal documents, send emails, and more.

One of the main benefits of this is that most Workspace subscriptions come with Workspace Studio. I’ve written multiple articles on this hidden AI Orchestration engine. My favorite is how I use it to find jobs, score them, and load them into my CRM. It is this very tool that gave me the hack I use to save tokens in my n8n workflows. You can do it too without spending a dime in most cases by following the steps outlined below.

AI Agent Placement Makes All The Difference

When building out n8n, make.com, or Super Easy CRM workflows, my immediate thought is to insert an AI Agent somewhere into the mix. However, this will immediately result in token use. And, if you scale your solutions or sell them to clients like I do, costs can and will climb… fast!

If you place your agent in a flow, then the Gemini API will need to be utilized. However, if you turn on Google Workspace Studio Flows, then you can have Gemini work for you without paying a single token.

Here’s the architecture I use.

Google Workspace Flows + n8n Hand-off

1

Step 1

Daily, at 8:00 AM, Gemini is invoked and instructed to surf the Internet and find contracting opportunities in my area. It is tasked with the below:

  • Locate companies with 25 to 150 users
  • In heavily regulated industries notorious for clunky processes like healthcare insurance, immigration services, etc.
  • Score them based on fit
  • Place the results in a structured JSON format
2

Step 2

n8n listens for this email based on the subject. Once located, it parses the JSON, preps the payload, and sends it into Super Easy CRM for the next step of outreach.

This approach happens with no additional costs incurred.

lead scoring with Gemini

Are The Free Models as Good As the Paid Ones?

You may be thinking the advantage of using the paid, token-based Gemini is that you’ll be getting a superior product. And in many cases that may be true, but not in this one. At the time of writing, Google Workspace Studio just uses whatever version of Gemini is current. You get more flexibility in this regard through the API; however, I was running the gemini-3-flash-preview model and received nearly the same output I did from the ‘stock’ Gemini in Workspace Studio.

In this specific use case, the free option was just as good, or better, than the paid. Send this to your CFO to become his favorite person.

How Can Workspace Studio Hand Off Information Outside of Email?

If you don’t want n8n snooping around your Gmail, I get it. You’ve got options other than turning a powerful bot loose on your precious inbox. Here are some other entry points I’ve used when clients prefer to keep their inboxes out of the loop:

Google Drive

Create a folder called ‘Contract Opportunities’, then house a spreadsheet with opportunities in each row.

Dedicated Gmail

Instead of using your personal email, use something like bots@mattflows.com or ai@mattflows.com. This will limit the scope of what n8n has access to immediately.

Send a Webhook

I put this as number three because, if you’re not a developer, it can be a bit challenging to set up. But in essence, what you would be doing is sending data directly from the Flow to n8n to start the automation. The roles are reversed here in that n8n is receiving information, not gathering it.

Gemini’s Tight Integration with Workspace Studio Makes Flows Affordable and Powerful

The seamless integration of Gemini into Google’s core business products is the best around, in my opinion. It doesn’t feel like Microsoft’s Copilot, a bolted-on afterthought that seems to appear at the worst possible moments. If you’re using Google to manage your business, I highly recommend turning on Workspace Studio. Create a flow and let Gemini handle business logic and information gathering for free. And then, whenever you need something more complex or need data extracted from large files, summon the Gemini API to make your flows much more token efficient. That’s exactly the approach I took building my drug tier classification engine for a healthcare formulary client.