Google Workspace Studio Beginners Guide: Build AI Agents with Step-by-Step Tutorials

Updated: 2026-04-08

Google Workspace Studio launched in December 2025 and I stumbled across it in early January 2026. I’ve been automating Google Workspace since the days it was called Google Apps for Business, though it hasn’t been my main focus lately. But the more I dig into Workspace Studio, the more I think it’s worth paying close attention to, particularly some of the features Google initially released and then pulled back. The possibilities are quite interesting.

For now, this post is focused on the basics. I’ll be updating it regularly as I learn more, especially the detailed tutorials further down. The goal is a natural progression: a gentle introduction if you’re just getting your bearings, and more detailed real-world examples as you go deeper. One place to bookmark, that is kept current.

Access: Google Workspace Studio is available on Business Starter, Standard, and Plus; Enterprise Standard and Plus; and Education editions. Visit studio.workspace.google.com to check whether your account has access. For admin configuration, see Access and Setup.

Just getting started? The quickest way in is to describe what you want to automate in natural language and let Gemini build the flow for you. Once you’ve tried that, the rest of the post will make a lot more sense.

What is Google Workspace Studio?

Google Workspace Studio is a no-code platform at studio.workspace.google.com that lets you create “flows”; intelligent agents that connect your Google Workspace apps and make decisions on your behalf.

The critical difference from traditional automation tools is that Workspace Studio is AI-first. You define the intent (like “manage my inbox for urgent client issues”), and Gemini handles the execution logic. The system can distinguish between spam and a critical client complaint without you programming every possible scenario.

Think of Workspace Studio as connective tissue between previously siloed applications. A Google Form response no longer sits idle in a spreadsheet waiting for you to notice it. Instead, it becomes a living event that triggers emails, schedules meetings, and updates CRMs automatically.

What You Need Before You Start

• A Google Workspace account on a supported edition (Business Starter, Standard, or Plus; Enterprise Standard or Plus; Education)
• Gemini enabled on your account – your administrator controls this in the Admin Console
• Smart features and Google Workspace smart features toggled ON in Gmail Settings > General – required for email-triggered flows to run
• A computer with a supported web browser – flow creation is not available on mobile

If your account doesn’t have access yet, see Access and Setup at the end of this guide.

Getting to Know the Workspace Studio Interface

Access Workspace Studio at https://studio.workspace.google.com. The interface is divided into four primary workspaces:

The Discover Page is your entry point. It features a Describe a task for Gemini input where you can type natural language requests like “Create an agent that saves invoices to Drive.” It also houses pre-built templates categorised by role (Sales, HR, Admin).

The Flow Builder is where you construct agents. It’s a vertical, linear timeline where you click the + button to insert steps. The right-hand sidebar provides configuration options for each step, including variable mapping and prompt engineering fields.

My Flows is your dashboard for managing existing agents. It provides a list view with toggle switches to instantly enable or pause flows, plus a “Last Run” status for health monitoring.

Activity & Logs is critical for debugging. You can drill down into specific runs to see exactly what data entered and exited each step. This is essential for diagnosing why a Decide step returned False when you expected True.

Create Your First Flow by Describing It

The fastest way to start is the Describe a task for Gemini bar on the Discover page. Type what you want to automate in natural language, and Gemini generates a complete flow – starter, steps, and variables already configured. You review, adjust if needed, and activate.

1. Go to Workspace Studio
   • Navigate to studio.workspace.google.com
2. Find the description field
   • On the Discover page, locate the Describe a task for Gemini field at the top of the page

3. Describe your automation in plain English
   • Be specific about the trigger and the outcome: describe what starts the flow and what you want to happen as a result
   • You don’t need to know step names or how the flow is structured – describe the behaviour you want

Examples that work well:

• “Send me a daily summary of my unread emails every morning at 8am”
• “Send me a Google Chat message 15 minutes before each calendar meeting”
• “When a new row is added to my Google Sheet, send me a notification in Chat“
• You’ll then have to manually choose the spreadsheet, sheet, etc. So this is an example of one that gets you started, but you need to make edits to complete.

Specificity improves results. A vague description gives Gemini less to work with. Here’s a complete prompt that produces the Morning Briefing Agent in the next section — you can use it verbatim:

“Every weekday at 8am, scan my unread emails from the previous day. List emails that need a reply today under ‘Action Required’. Group everything else under ‘FYI’. Ignore newsletters and automated notifications. Send the formatted summary to me in Google Chat.”

The more specific the description, the more accurately the generated flow matches your intent and the fewer manual adjustments you need to make.

4. Review the generated flow
   • Gemini produces a draft flow with starters, steps, and variables already configured
   • Check that the trigger matches your intent – confirm sender filters, schedule settings, and conditions
   • Verify variable mappings in each step

5. Test and activate
   • Click Test run to verify the flow works before turning it on
   • Click Turn on when you’re satisfied

Starting from a Template

The Discover page includes a library of pre-built flows organised by use case. Browse the categories, pick one that matches your scenario, and Workspace Studio gives you a working flow – steps, variables, and triggers already configured. You adjust the specifics (which folder, which sheet, which people) and activate.

Email boosters

• Get a daily summary of unread emails
• Notify me about emails from key people
• Label emails with action items
• Notify me about urgent emails
• Auto-add email attachments to Drive
• Star emails for follow-up
• Notify me about messages with a keyword

Better meetings

• Get pre-meeting briefs in Chat
• Send guests summaries and action items after meetings
• Send me summaries and action items after meetings
• Get meeting reminders in Chat
• Auto-create tasks for me from meeting transcripts
• Draft meeting follow-up emails with Gemini

Tasks and action items

• Auto-create tasks when I’m sent action items
• Create tasks when customers send requests
• Auto-create tasks when files are added to a folder
• Notify me when I’m mentioned in spaces
• Notify me about emails with action items
• Notify me when a file is added to a folder

Customer connections

• Draft email replies to questions using a doc and Gemini

Everyday essentials

• Get news headlines summarised daily
• Get AI news briefs daily

When to Use Templates, Plain English, or Scratch

Use a template when:

• A pre-built option closely matches your use case
• You want to be up and running in under two minutes
• You’d rather see how a working flow is structured before building your own

Use plain English when:

• You have a specific scenario in mind that no template covers
• You want a custom starting point to edit from
• You’re exploring what’s possible without yet knowing step names

Build from scratch when:

• You need precise control over triggers, filters, and variable mapping from the start
• The generated flow doesn’t match your intent after one or two refinements

Templates and plain English both produce complete flows you can customise and activate. Scratch gives you exact control from the outset. Most users start with a template or description, then switch to manual editing when they need precision.

Your First Flow from Scratch: A Personal Briefing Agent

The best way to understand the system is to build something simple but useful. We’ll use the morning briefing scenario from the previous section. If you already generated it using the plain English prompt, follow along to see what Gemini built and how each piece fits together. If you’re starting fresh, these steps build the same flow from scratch.

Scenario: You start your day with an overloaded inbox. Instead of manually checking your inbox first thing, this agent automatically triages your inbox while you sleep. At 8:00 AM, it identifies unread emails from a specific timeframe, categorises them by urgency and theme, and sends a beautifully formatted summary to your Google Chat.

Creating the Flow:

1. Access the Workspace Studio
   • Go to studio.workspace.google.com
2. Start from Scratch
   • On the Discover page, click + New Flow (or use the Describe a task for Gemini bar and type “Send me a daily summary of VIP emails in Chat”)
3. Name Your Agent
   • Click the pencil icon at the top left to rename your flow from “Untitled” to “Morning Briefing Bot”

Step 1: The Starter (The Trigger)

• Click Choose a starter and select On a schedule.
• Configuration:
• In the Start date and time section select a Date and set the Time time to 8:00am.
• For Repeat, select Every weekday (Monday to Friday) from the dropdown.
• You can also select when it Ends.

Step 2: The Intelligence (Recap unread emails)

• Click the Choose a step below your starter and select Recap unread emails step from the AI skills section.

• Configuration:
• Time range for summary
  • Select Yesterday (12 AM to 11:59 PM) from the dropdown
• Enter a prompt to generate a summary
  • This field comes with a sophisticated default prompt. This prompt is designed to:
  • Filter content: It automatically excludes ads, social media notifications, newsletters, and spam
  • Identify Action Items: It creates an “Urgent Action Items” category for emails requiring a direct response
  • Theme Grouping: It groups repetitive themes, such as project updates, into high-level summaries rather than listing every individual email
  • Customisation Tip:
  • You can edit this prompt to be more specific, such as adding “Focus specifically on emails from [Client Name]” to further refine the results

Step 3: The Delivery (Google Chat)

• Click + Add step.

• Select Notify me in Chat.

• Message Body:
• Click into the Message field and type the @ symbol (or click + Variables)
• Select the Variable:
• From the list, choose Step 2: Summary of unread email. It will appear as a coloured chip in your text

or

This passes the AI’s categorised summary into your chat message.

Activation and Testing

• Test Run:
• If you don’t want to wait until 8:00 AM, click Test run to execute the flow immediately and see the results in the Activity tab
• Note: Test runs execute the flow for real. This one sends an actual Google Chat message to you – that’s intentional, and it’s how you confirm the output looks correct before turning the flow on. For flows that modify files or send emails to others, test against dedicated test data first.

• Find Your Results:
• Your briefing will not appear as a standard DM from a person. Instead, look in the “Apps” section at the bottom of your Google Chat sidebar. You will see a new app named “Morning Briefing Bot” (the name of your flow)
• Turn On: Click the Turn on toggle

This simple three-step flow demonstrates the core loop: Trigger (Time) → Intelligence (Summarisation) → Action (Communication).

Understanding the Core Concepts

To use Workspace Studio effectively, you need to understand its specific architecture. The platform operates on four fundamental concepts: Starters, Steps, Variables, and the Intelligence Layer.

Starters: What Triggers Your Flows

Every flow begins with a single event called a Starter. This acts as an event listener waiting for a specific state change in your Google ecosystem. Currently, each flow can have only one starter; you can’t trigger the same flow from both an email and a chat message (you’d need two separate flows).

Starters fall into three categories:

Communication Events

Communication Events trigger from human interaction. When I get an email is the most common, hooking into Gmail with pre-processing filters for sender, subject, and attachments. This filtering is crucial, without it, your flow fires on every marketing newsletter, quickly exhausting your daily limits.

When I get a chat message and When I'm mentioned enable “ChatOps,” transforming Google Chat into a command-line interface for your business. When an emoji reaction is added might seem trivial, but it unlocks powerful approval workflows. A manager can approve a document simply by reacting with a ✅, which the system interprets as permission to proceed.

Data State Events

Data State Events monitor changes to static data. When a sheet changes turns Google Sheets into a database with triggers, monitoring specific columns or rows for updates. This is particularly powerful for status tracking, when a “Status” column changes to “Done,” your flow can automatically notify stakeholders.

When a file is edited in Drive detects content changes (but not metadata changes like renaming or moving, which prevents infinite loops).

Temporal Events

Temporal Events trigger based on time. On a schedule works like a cron job for batch processing. Based on a meeting is unique – it carries event metadata (attendees, description, attachments) into your flow, enabling contextual preparation or follow-up actions.

Steps: The Execution Chain

Once a Starter fires, your flow executes a sequence of Steps. You’re currently limited to 20 steps, which enforces modular design over monolithic complexity.

Action Steps are deterministic operations that interact with APIs: Send email, Create Drive Folder, Add row to Sheet, or Post Message to Chat.

Transformation Steps manipulate data in transit. The Extract step uses AI to parse unstructured text (like an invoice) into structured data (Invoice ID, Amount, Date).

Reasoning Steps differentiate Workspace Studio from traditional automation. The Decide step doesn’t just compare values, it evaluates semantic meaning. You can ask it to determine whether an email comes from a frustrated customer likely to churn. The Ask Gemini step allows for sophisticated prompt engineering within your flow, letting you guide exactly how the AI responds.

Variables: Your Data Pipeline

Data flows through your agent via Variables. When a Starter executes, it generates an object containing relevant metadata. A “New Email” starter generates Sender, Subject, Body, Date, and Attachments.

Subsequent steps can reference these variables. Your reply draft might use the Sender variable to address the recipient and a Summary variable from an AI summarisation step to generate content.

In the Workspace Studio interface, variables appear as chips you can insert into text fields. Understanding data types is critical here, a File variable (an object reference) can’t be placed into a Text field without causing errors or producing useless object ID strings.

The Intelligence Layer: Gemini

Google’s Gemini model underpins the entire system. Unlike traditional code that produces identical output for identical input, Gemini’s output can vary slightly in phrasing or tone. This probabilistic nature is a feature, allowing for creativity and adaptability in communication workflows.

This requires a mindset shift. You need to design flows that are resilient to slight variations in AI output, often by using the Decide step to validate responses before taking action.

Key Capabilities That Set Workspace Studio Apart

Advanced Event Monitoring

The Starter mechanism is more sophisticated than simple triggers. It includes pre-flight filtering that’s essential for quota management.

Granular Email Filtering doesn’t just fire on “New Email.” You can configure Boolean logic at the trigger level: Subject CONTAINS "Invoice" AND HasAttachment == TRUE AND SenderDomain != "internal.com". Setting this up correctly is the single most important step in preventing runaway automation that exhausts your daily limits.

File Event Nuance distinguishes between files being added versus edited. The File Edited starter detects content changes but currently doesn’t trigger on metadata changes like moving (though this feature in Google Drive is currently in developer preview). However, simply renaming a file still generally does not trigger a Content Edited flow.

Calendar Context with the Based on a meeting starter allows for relative time triggers (15 minutes before, 30 minutes after). It carries event metadata into your flow, enabling highly contextual preparation or follow-up actions.

The AI Reasoning Engine

Gemini integration enables steps that were previously impossible in automation.

The Decide Step is a probabilistic logic gate. Unlike standard conditional checks that compare strings, Decide evaluates sentiment and intent. You can ask it: “Is this email from a frustrated customer who is likely to churn?” The system feeds the email body and your prompt to Gemini, which returns a Boolean to guide your flow. This enables what you might call “vibe-based” routing.

The Extract Step uses Gemini’s Named Entity Recognition capabilities. It can ingest a PDF, image, or messy email thread and extract specific entities you define (Tracking Number, SKU, Meeting Date). It creates structured variables from these entities, essentially cleaning messy human data for machine processing.

Prompt Engineering in Ask Gemini gives you a blank canvas. You can use sophisticated prompting techniques (Few-Shot Prompting, Chain-of-Thought) to guide the AI’s output. For example, you might provide three examples of your desired writing style to ensure drafted emails match your voice. One configuration detail that significantly affects output quality: in the Sources field, select All sources rather than leaving it on the default. This allows Gemini to search your actual Gmail and Drive content, not just respond from its training data. Without this, Ask Gemini steps produce generic answers instead of context pulled from your real emails and documents.

Native Application Integration

Workspace Studio has superuser access to Google Apps, often bypassing the limitations of public APIs.

Gmail flows can manage labels, mark messages as read or unread, and draft replies in your Drafts folder (a best practice for human review before sending).

Drive flows can manage folder hierarchies, move files, and control permissions. The file iterator capability (processing all files in a folder) is particularly useful for batch organisation.

Chat flows can interact with specific Spaces and distinguish between threaded replies and new conversations, enabling organised automated discourse.

Sheets flows can use Update Row and Find Row steps to function as a lightweight relational database. You can look up a customer ID in Column A and update Status in Column F, effectively turning Sheets into a CRM backend.

Connectivity and Extensibility (Currently in Limited Preview)

Limited Preview. There is currently no way to request access to the limited preview

Workspace Studio connects to the external world through two primary mechanisms.

Webhooks: Your Digital Messenger

Think of a webhook as a way for your flow to send a specific message to another app. A webhook step sends an HTTP request to a URL you provide, allowing you to trigger actions in external services.

• How it works: You provide a static URL for the external service and choose a “method” (like POST to create something new or GET to retrieve data).
• The Payload: This is the actual information you are sending, such as a customer’s name or an email summary. We recommend using JSON format for this, as it is the standard language web services use to talk to each other.
• Example: You could set up a flow that, upon receiving an “Urgent” email, automatically sends a webhook to Slack to post a message in an emergency channel

Third-Party Integrations: Ready-to-Use Connections

For those who don’t want to deal with URLs and payloads, Workspace Studio offers pre-built integrations. These allow you to drag and drop actions from popular business apps directly into your execution chain.

• Common Connectors: You can currently find integrations for Asana, Jira, Mailchimp, and Salesforce.
• Simple Setup: These integrations handle the technical “handshaking” (OAuth authentication) for you. You simply connect your account through the Google Workspace Marketplace and follow the on-screen instructions.
• Beginner Tip: Always make sure you trust the third-party service before connecting it, as your flows may share sensitive Google Account data, like Gmail message contents or Calendar event details, with that service

Add-on Extensions: Custom Power with Apps Script

If a pre-built step doesn’t exist for your specific needs, you (or a developer in your organisation) can build one using Google Apps Script.

• Custom Steps: These are tasks tailored to your specific business logic that can be shared across your organisation.
• Complex Data Handling: For more advanced users, extensions allow for Dynamic Variables (information that changes based on what you select during setup, like specific questions in a Google Form) and Custom Resources (which group multiple related variables together, like a customer’s name, email, and address)

Advanced AI: Ask a Gem

Extensibility also includes specialised AI agents called Gems. In the Ask a Gem step, you can use an AI agent that has been pre-trained for a very specific expertise, such as a “Brainstormer” or a “Sales Pitch Ideator,” rather than just asking general questions to Gemini

Building AI Agents: Detailed Tutorials

Tutorial 1: The Dynamic Escalation Router

Scenario: Your team flags urgent issues in Google Chat, but the message gets buried. You then have to manually figure out who needs to know, write the escalation email, and log it somewhere. Three manual tasks from a single emoji.

Objective: React to any Chat message with 🚨 and the flow handles the rest – it extracts the issue, category, and urgency from the message, looks up the right contact from a routing spreadsheet, pre-fills a formatted escalation email draft addressed to that contact, and logs the incident to your private tracker.

Before you build: Create a private Google Sheet (it can’t be shared with others – see Limitations) and call it Escalation Hub say, with two tabs:

• Escalation Roster – two columns: Category (e.g. Billing, Support, Sales) and Contact Email
• Escalation Log – columns: Date, Sender, Issue, Urgency, Category

Construction:

Starter

Step 1: When an emoji reaction is added

• Choose the police-car-light emoji: 🚨
• Ensure All messages is selected for Conversation type.

• Note: This starter provides Message Text, Message Sender, and Space Name as variables.

Actions

Step 2: Extract

Add the Extract step to your flow. This step uses Gemini to read unstructured text and pull out specific data points.

• Action: Extract
• Input: {Message Text}
• Fields to extract: Issue Summary (Text), Urgency Level (Text), Issue Category (Text – matching the categories in your Escalation Roster, e.g. “Billing”, “Support”, “Sales”)
• Gemini reads the raw message and returns three structured variables you can use in the steps below.

1. Content to analyze
   • Click + Variables (or type @) and select the {Message text} variable from your When an emoji reaction is added starter.

2. What to extract
   • Item 1
     • Custom content name: Issue Summary
     • Description for Gemini: A brief, 1-2 sentence summary of the core problem or request.
   • Item 2
     • Custom content name: Urgency Level
     • Description for Gemini: The urgency of the issue based on the context. Return only "High", "Medium", or "Low".
   • Item 3
     • Custom content name: Issue Category
     • Description for Gemini: The department this issue belongs to. You must return EXACTLY one of these three words: "Support", "Billing", or "Sales".

Why this matters: By giving Gemini strict instructions in the description for “Issue Category”, you ensure it outputs a word that perfectly matches the columns in your Escalation Roster spreadsheet. If Gemini outputs “Technical Support” instead of exactly “Support”, the spreadsheet lookup in the next step will fail!

Step 3: Get sheet contents

• Action: Get sheet contents
• Sheet: your private Escalation Roster
• Filter: Category column matches {Issue Category}
• Output: [Matching values "Contact Email"] – the email address for whoever owns that category

Step 4: Draft an email

• Action: Draft an email
• To:
  • Step 3 > {Matching values "Contact Email"}.
• Subject:
  • Step 2 > {Issue Summary}.
• Body:

Flagged by: {Sender's full name}
Urgency: {Urgency Level}

Original message:
{Message text}

Please review and action.

The draft lands in your Gmail Drafts folder, addressed to the right person, formatted and ready to send.

Step 5: Add a row

• Action: Add a row
• Sheet: your private Escalation Log

• Mapping:
• Date
  • Step 1 > {Post time}
• Sender
  • Step 1 > {Sender's email address}
• Issue
  • Step 2 > {Issue Summary}
• Urgency
  • Step 2 > {Urgency Level}
• Category
  • Step 2 > {Issue Category}

Outcome: One emoji reaction extracts the issue, routes it to the right person, and logs it – without opening another tab. The mental model this flow demonstrates – extract unstructured data with AI, query a spreadsheet for conditional routing, act on the result – applies to dozens of other workflows once you’ve built it once.

Tutorial 2: The Invoice Bot

Scenario: Vendors send PDF invoices to your email. You manually save them to Drive and log amounts in a Sheet.

Objective: Automate the extraction and logging of invoice data.

Construction:

Starter

Step 1: When I get an email

• Click Choose a starter > When I get an email.
• Click the Specific emails button.
• In the Has the words field, enter invoice OR bill OR receipt OR statement.
• Click Show more to see more options.
• Has your email address in Cs or Bcc should be selected by default.
• Enable the Has attachment checkbox.

Actions

Step 2: Filter a list of email attachments

• Click Choose a step > Filter.
• In Apply filter to click + Variables > Step 1: Email attachments.
• For the If condition select File type and then is should populate by default in the other dropdown (ensure that it is the selection if not).
• Select PDFs from the dropdown below the other two.

Step 3: Add email attachments to Drive

• Click + Add step > Add email attachments to Drive.
• In the Location for email attachments section, click the Drive button and select a folder to deposit the pdfs.
• In the Gmail attachments section click + Variables > Step 2: Filtered items > {Filtered items}.

The Add email attachments to Drive step is the first step to get access to the data in the attachment. We must therefore have it before the Extract step.

Step 4: Extract

• Click + Add step > Extract.
• For Content click + Variables > Step 3: Add email attachments to ... > {Links to the files}.
• What to extract
  • Item 1
    • Custom content name: enter Vendor Name.
    • Description for Gemini: enter The legal name of the company or person issuing the invoice
  • Item 2
    • Custom content name: enter Invoice Number
    • Description for Gemini: enter The unique identification number assigned by the sender to this specific bill
  • Item 3
    • Custom content name: enter Total Amount
    • Description for Gemini: enter `The final total balance due including all taxes and fees
  • Item 4
    • Custom content name: enter Due Date
    • Description for Gemini: enter The deadline by which the payment must be received

Step 5 Get sheet contents

• click Choose a step > Get sheet contents.
• To Select the sheets to search under Spreadsheet click + Spreadsheet > {Accounts Payble Ledger} or whatever is the Google Sheet you are using.
• Choose the Sheet has your invoices from the dropdown, in this example in is Invoices.
• In the Find rows to get values from section:
  • For Column choose Vendor Name then for Value click + > Step 4: Extract > {Vendor Name}.
  • Click And.
  • For Column choose Invoice Number then for Value click + > Step 4: Extract > {Invoice Number}.

Step 6: Check if Step 5: Number of matching rows > 0

• Step 7: Notify me in Chat
  • Click Add substep > Notify me in Chat.
  • Message: input 🛑 Duplicate invoice# {Step 4: Invoice Number} detected from {Step 4: Vendor Name}.
• Step 8: Add labels
  • Click Add substep > `Add labels`.
  • If you don’t already have a label, under Your labels click + New Label enter Duplicate Invoices - Review for the New label name > Create.
• Step 9: Archive an email
  • Add substep > Archive an email > under Email to archive + Variables > Step 8: Add labels > Email ID (ie {Step 8: Email ID}).

Step 10: Check if Step 5: Number of matching rows is 0

• Click + Add step > Check if.
• If > {Step 5: Number of matching rows} > is > 0.
• Step 11: Add a row
  • Click Add substep > Add a row.
  • To Select the sheets to search under Spreadsheet click + Spreadsheet > {Accounts Payble Ledger} or whatever is the Google Sheet you are using.
  • Choose the Sheet has your invoices from the dropdown, in this example in is Invoices.
  • Add data by column
    • Vendor Name
      • Use @ or + Variables > Step 4: Extract > {Vendor Name}
    • Invoice Number
      • Use @ or + Variables > Step 4: Extract > {Invoice Number}
    • Total Amount
      • Total Amount to + Variables > Step 4: Extract > {Total Amount}
    • Due Date
      • Use @ or + Variables > Step 4: Extract > {Due Date}
    • Invoice Link
      • Use @ or + Variables > Step 3: Add email attachments to ... >{Links to the files}.
    • “Status
      • Enter text Review Needed

Outcome: A fully automated Accounts Payable ingestion engine.

Tutorial 3: The Sales Lead Route

🚨 Important Note: This tutorial uses Webhooks and Third-Party Integrations. These features are currently in Limited Preview. If you do not see these options in your Workspace Studio environment, your account does not yet have access to this experimental track.

Scenario: A “Contact Us” form on your website collects leads. You want to route high-value leads to Salesforce and low-value leads to an email nurture sequence.

Construction:

Starter

Step 1: When a form response comes in

• Click Choose a starter > When a form response comes in.
• Select the form you want to monitor: Choose your “Contact Us” form from the dropdown.

Actions
Step 2: Decide
AI Qualification

• Click Choose a step > Decide.
• In the Enter a prompt text box, write the question you want Gemini to answer with a true or false decision:
  Analyse this lead inquiry: {+ Variables > Step 1: Form Response}. Is this a high-value enterprise lead (asking for bulk pricing, 100+ seats, or from a Fortune 500 domain)? Return TRUE for Enterprise, FALSE for SMB.
• Note: A Decide step is always paired with a Check if step that is prefilled to check if the Decide result is true.

Step 3: Check if Step 2: Decision is true
Route High-Value Leads (If True)

• Inside the automatically generated Check if step, add your integration step.
• Select your Salesforce integration connector.
• Map the data using your variables from Step 1.

Step 4: Create the "Else" Branch
Create the “Else” Branch

• To have the flow do steps for the opposite condition, add another Check if step with the matching set to the opposite of the first.
• Set the condition to check if the Decide result is false.

Step 5: Route SMB Leads (If False)

• Inside this second Check if step, click + Add step and select Send a webhook.
• Webhook or API URL: https://api.mailchimp.com/3.0/lists/....
• Method: Choose the HTTP method for your request, such as POST to send data.
• Payload: Enter the content for the request body using your JSON payload. You can use a variable to add content to the payload:

{
"email": "{+ Variables > Step 1: Email}",
"tags": ["nurture_sequence"]
}

Outcome: Automated lead qualification that prioritises high-value prospects instantly.

Why Google Released This Now

By late 2025, employees were drowning in disparate AI tools. The constant switching between chatbot windows, email clients, and document editors created what’s known as “toggle tax” – the mental cost of context switching.

Google’s strategy with Workspace Studio is consolidation. By embedding intelligent automation inside your existing workspace, they address the two biggest barriers to enterprise AI adoption: context switching and data security.

There’s another factor at play. Employees were increasingly using unapproved tools like Zapier or personal ChatGPT accounts to automate their work. This “Shadow IT” posed serious risks to data governance. Workspace Studio brings automation activity back under organisational control, allowing administrators to monitor, audit, and manage how AI interacts with corporate data.

How Workspace Studio Compares to Other Automation Tools

To understand where Workspace Studio fits, you need to see it in the context of the broader automation market dominated by Zapier, Make, n8n, and Zenphi.

The Cost and Security Advantage

For existing Google Workspace Business and Enterprise customers, Workspace Studio is largely included in your subscription. Compare this to Zapier or Make, which charge by the task or operation. A company running 10,000 automations monthly could pay hundreds to Zapier – with Workspace Studio, this cost is absorbed into your seat licence (subject to daily limits).

Third-party tools like Zapier require your data to leave the Google tenant. An email arrives in Gmail, gets sent to Zapier’s servers for processing, then action is taken. This data egress creates security vulnerabilities and compliance headaches (GDPR, HIPAA). Workspace Studio processes data within the Google compliance boundary; a massive advantage for regulated industries.

Competitor Comparison

Here’s how Workspace Studio compares to the main alternatives:

Feature	Google Workspace Studio	n8n	Zenphi	Zapier
Primary User	Business Analyst / Generalist	Developer / Technical Ops	Enterprise Process Architect	SMB / Generalist
Architecture	SaaS (Google Cloud)	Self-Hosted or Cloud	SaaS (Google Cloud)	SaaS
Logic Model	Agentic (AI/Probabilistic)	Deterministic (Node-based)	Process-Oriented (BPMN)	Deterministic (Trigger-Action)
Data Sovereignty	High (Stays in Google)	Highest (Self-Hosted)	High (Google Native)	Low (Third-Party Processing)
Complexity Limit	Low (20 Steps, No Loops)	High (Infinite Complexity)	High (State Machines)	Medium (Multi-step Zaps)
Pricing	Included in Workspace*	Licensing / Free Community	Subscription	Usage-Based (Task count)

n8n remains superior for complex, technical orchestration. If you need to transform massive JSON arrays, run Python scripts, or connect to on-premise SQL databases via VPN, n8n is the tool. Workspace Studio can’t match n8n’s flexibility or self-hosting capability.

Zenphi positions itself as a “Business Process Automation” tool rather than a “Task Automation” tool. It handles long-running state (like a document approval process taking three weeks and involving five people). Workspace Studio flows are transient, they run and finish. Zenphi also offers deeper admin governance features that Workspace Studio currently lacks.

Zapier remains the king of connectivity. With 6,000+ apps, if you need to connect to a niche marketing tool or specific project management software not supported by Google’s limited Marketplace, Zapier is still necessary.

Understanding the Limitations

Google is providing promotional access to higher usage limits of Workspace Studio to allow users to experiment with these features. However, usage may be subject to change to ensure consistent performance for all users.

The following are the limits as of early 2026:

Capacity and Quotas

• The 20-Step Ceiling means a single flow cannot exceed 20 steps. This forces you to break complex processes into smaller, daisy-chained flows, which can make debugging difficult.
• Flow Creation Cap: You can create a maximum of 100 total flows per account, which includes both active and inactive (turned off) agents.
• Gmail Starter Restrictions: To manage high-frequency events, a maximum of 25 active flows can be triggered by Gmail starters at any one time.
• Daily Run Limit: There is a combined maximum number of times all of your flows can run within a 24-hour period. If this limit is reached, all active flows will stop running until the limit resets.
• Gemini Prompt Limits: When using AI steps like “Ask Gemini” or “Decide”, your total prompt must remain under approximately 144,000 characters (roughly 18,000 words). This limit is cumulative, meaning it includes your manual instructions, all content added via variables (such as a “Message Body”), and any linked content. If a flow attempts to process a very long email thread or a massive document that pushes the prompt over this ceiling, the run will fail with an error. To avoid this, consider using fewer variables or ensuring the source data is concise.

Performance

• Latency is real for AI steps. A Decide step involves an LLM inference call, which can take 2-5 seconds. Workspace Studio isn’t suitable for real-time, sub-second latency requirements.

Logic and Conditional Constraints

Workspace Studio’s linear design results in specific limitations for branching logic:

• Variable Isolation: You cannot use variables generated within a conditional substep (like an “Ask Gemini” result inside a “Check if” branch) in later steps of the main flow.
• No Nesting: Flows cannot have nested conditional steps. You cannot place a “Check if” or “Decide” step inside the substeps of another conditional block.
• Operator Uniformity: When setting multiple conditions in a “Check if” step, you must choose either AND or OR; you cannot mix these operators within a single step.
• No Native Loops means you can’t iterate over a list of items within a flow. If a sheet change provides 10 modified rows, the flow processes them as a batch or processes only the first one, depending on configuration. You can’t say “For Each Row, Send Email”; a significant limitation compared to Apps Script or n8n.

Identity and Context Limitations

• The User Context Problem means flows execute with the permissions of the flow owner. If the owner loses access to a shared folder, the flow fails. There’s currently no concept of a “System Flow” that runs with elevated, independent privileges. This links automation fragility to employee turnover.

• Approval Workflow Gaps are significant. The platform lacks a built-in “Wait for Approval” state. You can’t pause a flow for two days while a manager reviews a document. You must construct workarounds using two separate flows (Request Flow and Approval Response Flow), which is cumbersome compared to Zenphi’s native approval steps.

Security and Access Limitations

• Age Restrictions: Users under 18 on school accounts are strictly prohibited from using any AI features; AI steps will be automatically removed if a flow is shared with them.
• Device Requirements: You must use a computer with a supported web browser to create or edit flows. Workspace Studio is not available for flow creation on mobile devices.
• File Permissions: Flows may fail to update files if the file is shared with other people or groups; it is recommended to use files that only you can access for automated updates.

Users under 18 (in K-12 Education domains) are blocked from using AI features in Workspace Studio, significantly limiting education sector use cases.

Best Practices for Production Deployment

Governance and Security

Administrators must actively manage the environment. This means:

• Setting clear policies on what types of automation are permitted
• Regularly auditing active flows to prevent data sprawl
• Establishing naming conventions for flows to make them discoverable
• Creating documentation standards so flows can be maintained when employees leave

Education and Training

Users need training not just on how to build flows, but when to use AI versus standard logic. Some decisions benefit from AI’s flexibility, while others require the predictability of deterministic rules.

Create internal champions who can help colleagues design flows effectively. The democratisation of automation doesn’t mean everyone should automate everything; it means the right people should automate the right things.

Strategic Positioning

Define where Workspace Studio fits in your broader automation stack. For simple, Google-native workflows with AI requirements, Studio is ideal. For complex, multi-system orchestration or long-running approval processes, you may need Zenphi or n8n alongside Studio.

Don’t force Workspace Studio into use cases where its limitations create more problems than it solves. The 20-step limit and lack of loops make certain workflows genuinely difficult.

Access and Setup

Supported Editions

Google Workspace Studio is available on the following editions:

• Google Workspace: Business Starter, Standard, and Plus; Enterprise Standard and Plus; Education Fundamentals, Standard, Plus, and the Teaching and Learning add-on
• Google AI: Google AI Pro for Education; Google AI Ultra for Business

Release Tracks (fully rolled out March 2026)

Google distributes new features through two tracks. Your organisation’s administrator chooses which track to use.

Rapid Release – Workspace Studio became available on this track from December 2025. Users on Rapid Release see new features as soon as they roll out from Google.

Scheduled Release – Workspace Studio became available on this track from February 2026. New features arrive at least one week after Rapid Release, giving organisations more time to prepare.

Workspace Studio is now fully rolled out on both tracks as of the second half of March 2026.

Administrative Setup

Before you can create your first flow, your Google Workspace Administrator must ensure several requirements are met in the Admin Console:

Workspace Studio is a core service and will be on or off by default based on your organization’s release track setting.

• To Turn On Workspace Studio:
• Navigate to Main Menu > Apps > Google Workspace > Workspace Studio.
• Click Service status.
• To turn a service on for everyone in your organisation, click On for everyone and then click Save.
• Alternatively, enable it for specific organisational units or groups.
• Enable Gemini Access:
• To use any AI-powered steps, such as “Ask Gemini” or “Recap,” your admin must explicitly enable Gemini on your account.
• Marketplace & Integration Setup:
• If you intend to use third-party connectors (like Salesforce or Jira), your admin may need to allow these integrations in the Google Workspace Marketplace or install a required helper application for specific services.
• Mandatory Gmail Settings:
• Even if the admin has enabled everything, your flows will fail to start if you have disabled smart features. You must go to your Gmail Settings > General and ensure both Smart features and Google Workspace smart features are toggled ON.
• Platform Restriction:
• You must use a computer with a supported web browser to create or edit flows; Workspace Studio is currently not available on mobile devices for the creation process.

Gemini Alpha Features

If your administrator wants access to experimental Gemini features beyond what’s available by default, they can enable the Gemini Alpha programme.

Important: The Alpha programme gives access to ALL experimental Gemini features, not just Workspace Studio. This means users will see other pre-release features that are still in development.

How your administrator can enable Alpha access:

As of April 2025, all Google Workspace customers can access Gemini alpha features without needing a specific add-on.

1. Sign in to the Google Admin console with an administrator account.
2. Navigate to Menu > Generative AI > Gemini for Workspace (Note: This requires Gemini Settings admin privileges).
3. Click on the Alpha Gemini features panel.
4. On the left, choose to enable access for specific Users, Groups, or Organisational units.
5. Click On and follow the prompts to confirm.

What to know about Alpha access:

• All-or-Nothing:
  You cannot pick and choose which alpha features to enable, it’s all of them or none.
• Data Security:
  Alpha features follow the same data protection standards as core Google Workspace services. Your data is not used to train Google’s general AI models.
• Feature Availability:
  Some advanced steps (third-party connectors, webhooks, etc) may be in “Limited Preview” or temporarily unavailable as Google scales infrastructure.

Next Steps

If you’re ready to implement Google Workspace Studio, start with these immediate actions:

Verify your organisation has Workspace Studio enabled. Check with your Google Workspace Administrator whether the service is turned on for your Organisational Unit. If not, request activation.

Start with a personal use case. Don’t begin by automating critical business processes. Build a simple flow that solves your own problem; like the morning briefing agent or automated file organisation. This builds your understanding without risking business operations.

Map your most time-consuming manual tasks. Spend a day noting every time you manually move data between systems or perform repetitive categorisation. These are your highest-value automation targets.

Build with the limitations in mind. Design flows that stay within the 20-step limit. If you need more complexity, plan how you’ll chain multiple flows together. Accept that some workflows genuinely aren’t suitable for Workspace Studio yet.

Google Workspace Studio represents a significant capability, but it’s a tool that requires thoughtful implementation. Success comes from understanding both its strengths and its constraints, then applying it strategically where it genuinely improves productivity.

Conclusion

The release of Google Workspace Studio marks the moment when AI agents transitioned from theoretical concept to practical utility in everyday work. For organisations, it presents a genuine opportunity to reclaim lost productivity, the hours spent moving data from email to sheets, from Drive to Chat, can now be handled by Gemini.

But Workspace Studio isn’t a magic solution. It’s a tool that requires governance, education, and strategy. Administrators must manage the environment to prevent security risks. Users must understand when to use AI versus standard logic. Organisations must define where Studio fits alongside other automation tools in their stack.

As we move deeper into 2026, Google will likely address current limitations like loops and service accounts. But the capabilities available today are already sufficient to transform personal and team productivity. The question isn’t whether to use Workspace Studio; it’s how to use it strategically, with clear understanding of both its power and its constraints.

Frequently Asked Questions