
We've all been there, where you have 60 seconds before your next meeting starts. You're searching...
We've all been there, where you have 60 seconds before your next meeting starts. You're searching your inbox for that one PDF a client sent three days ago, or trying to remember what was discussed in last week's thread.
In the past, you needed a human Chief of Staff or a complex web of Python scripts to solve this. Today, you just need Google Workspace Studio.
In this tutorial, we are going to build a Meeting Prep Agent. This AI agent will automatically wake up 15 minutes before any meeting on your calendar, scan your recent emails and documents for context, and send a concise Google Doc directly to your Google Chat.
If you haven't read my previous tutorial on building your first AI agent in Workspace Studio, start there for the basics. Prefer to watch instead of read? Check out the video version of that guide here.
Prefer to watch instead of read? Check out the video version of the guide here.
{% embed https://youtu.be/onNH6yJNtMs?si=8-h_8fMjkAdYa0TV %}
Google Workspace Studio is a no-code AI agent builder launched in December 2025. It lets anyone describe a workflow in plain English, and Gemini builds the automation flow for you.
Agents run securely inside your Google Workspace, respecting all your organization's permissions and data policies. No Apps Script, no third-party tools - just natural language and seamless integration with Calendar, Gmail, Drive, and Chat.
Key Features Relevant Here:
Prerequisites
To build this, we'll use a specific four-part logic within the Workspace Studio editor:

Navigate to studio.workspace.google.com and create a new flow. Instead of using the Describe to Design box, we'll build this manually to ensure precision.
Click on Choose a Starter and select Based on a meeting.


Once done, go ahead and click on Choose a step.
Now we need the agent to find the context. Add a new step and select Ask Gemini.

This opens the configuration window where you can talk to the AI in natural language. This is where you tell Gemini exactly what to look for. Because this agent is grounded in your Workspace, it can read your history to prepare you.

In the Enter a prompt box, copy and paste this structure:
I have an upcoming meeting. Here are the details: Title: [Step 1: Meeting title] Description: [Step 1: Event description] Attendees: [Step 1: All guest full names] Please search my recent emails with these attendees and any related Drive files to create a concise meeting briefing. Include key objectives, recent discussion points, and any action items I should be aware of.
Don't just type the bracketed text! Click the + Variables button in the bottom right of the prompt box. This allows you to select the Blue Chips from Step 1.
By mapping the Meeting Title, Description, and Attendee Names, you are giving Gemini the specific search terms it needs to scan your history.

Look at the bottom of the Ask Gemini panel. Under Sources Gemini can use, ensure All sources is selected.
Web Search: This allows Gemini to look up recent public news about the attendees or their companies.
Workspace: This is the magic part - it allows Gemini to securely read your internal emails and Drive files to find those specific action items mentioned in the prompt.

Once Gemini has finished its research, we want that information saved somewhere accessible. We'll use the Create a doc action to generate a dedicated briefing file in Google Drive.


The last step is the delivery. Since you're likely on the move or prepping between calls, a Google Chat notification is the fastest way to get your briefing.

In the Message box, we are going to combine text with variables from all three previous steps to create a high-value notification:
By including the link from Step 3, you can jump from your Chat app directly into the formatted Google Doc if you need more details during the meeting.

With your logic built, it's time to move from Design to Deployment. Google Workspace Studio offers a powerful way to verify your agent before it goes live.
The Test Run (Verification) Before you commit to the automation, click the Test run button at the bottom of the editor. This opens a configuration panel on the right.

Workspace Studio provides real-time feedback. As the agent runs, you'll see green checkmarks appear next to each step.


Going Live (Turn On) Once you are happy with the test results, click the blue Turn on button. Within a few seconds, you'll see a confirmation snackbar at the bottom left:
Your flow is on and ready to work!
Your AI Meeting Assistant is now officially on duty. It will sit silently in the background and trigger automatically 15 minutes before every meeting on your calendar.

The true power of this agent is visible in the output it creates for you. Within seconds of the trigger, two things happen:
You receive a notification from the AI Meeting Assistant app. It's not just a reminder; it's a full executive summary. It lists the date, time, attendees with their email links, and the core purpose of the meeting. It even includes a Note on Action Items - even if none were found, it gives you the peace of mind that it actually checked.

Simultaneously, a new document titled Meeting Brief: Project Quantum - EMEA Infrastructure Pivot Review is created in your Drive. This is a perfectly formatted document containing all the context Gemini gathered. You can open this doc during your call to take notes, knowing your background research is already done and documented.

By building this Meeting Prep Agent, you've done more than just automate a few clicks; you've fundamentally changed your relationship with your calendar. Instead of spending your day searching for context, you are now spending your day acting on it.
This workflow is a Force Multiplier. While you are focused on your current task, your AI agent is already working 15 minutes into the future, ensuring you walk into your next conversation with total confidence.
The "Meeting Prep" agent is just the beginning. You can now take these same principles - Starters, Variables, and Actions - and apply them to every corner of your work.
gemmaI ported the whole Gemma-4 family — E2B, E4B, 12B, 31B, and the 26B-A4B MoE — to run on...
communityHey DEV, I'm Tobore. Let's actually connect. I've been on here for a while now, mostly writing and...
ai(yep, kinda clickbait, just for the funsies 😊) At the beginning of the year, I relaunched my...
aiMy laptop was sitting idle with the fan at full tilt. Nothing was running that I knew of. The culprit...
githubactionsI Built a Thing! TL;DR — Google Gemini-based Pull Request reviews and Issue Triaging for...
aiI've been hearing the word "harness" thrown around a lot lately. I assumed it just meant "the IDE" or...
Workflows from the Neura Market marketplace related to this DeepSeek resource