Google’s new “agents” for Workspace - NOW AVAILABLE FOR EVERYONE · AI for LinkedIn

5d • Ideas

Google’s new “agents” for Workspace - NOW AVAILABLE FOR EVERYONE

Google’s new “agents” for Workspace are Gemini-powered automations you can design to read, reason over, and act on your Gmail, Drive, Docs, Sheets, Calendar, and more, all tied to your Google Workspace account and org settings.

What Google Workspace agents are:

Google is rolling out a hub called Google Workspace Studio where you can create, manage, and share AI agents that automate everything from email triage to multi-step approval workflows, without writing code. These agents run “inside” Workspace, so they can see relevant messages, files, and context and then take actions like labeling emails, drafting replies, updating docs/sheets, posting to Chat, or pinging humans when needed.

Under the hood, the agents use the Gemini 3 model (successor to Duet AI) so they can reason over longer, messier workflows rather than just doing single prompts like “summarize this email.” You describe the outcome in natural language (e.g., “Whenever a customer email mentions ‘refund’ and includes an invoice, route it to Support, log it in a Sheet, and draft a reply”) and Studio turns that into a runnable agent.

How agents connect to Workspace accounts:

Workspace agents are tied to your Google Workspace domain and user accounts, not to personal @gmail.com accounts. Admins decide who in the organization can use Workspace Studio, which apps agents can access (Gmail, Drive, Chat, Calendar, etc.), and what data boundaries apply, so the agents follow the same security, sharing, and compliance rules as your normal Workspace content.

From a user’s point of view, agents show up where you already work: in Gmail side panels, Chat, Docs, Sheets, and as automations that run in the background when triggers fire (like receiving an email or hitting a time/date condition). Because they’re Workspace-native, agents can respect things like document permissions, labels, and org policies when reading or acting on data.

Key building blocks: triggers, steps, and variables:

Workspace Studio gives you three main pieces when designing an agent: Starters (triggers), Steps, and Variables.

Starters define when the agent runs: for example, “when an email arrives to this alias,” “at 8am daily,” or “when a form is submitted.”

Steps are the actions: analyze an email with Gemini, extract entities, draft a response, create or update a Doc/Sheet, send a Chat message, call a webhook, etc.

Variables act as placeholders for dynamic data, like “sender email,” “Gemini’s classification result,” “invoice amount,” or “summary text,” which later steps can reuse.

Because Gemini handles the reasoning layer, agents can do things like detect which emails actually contain questions, pull out action items or invoice numbers from attachments, and then route or summarize them appropriately.

Licensing, editions, and Gemini add-ons:

For now, Workspace Studio and agents are rolling out to Business and Enterprise Workspace tiers, and you generally need a Gemini for Workspace or related add-on (what used to be branded as Duet AI) to unlock the AI capabilities. Google positions “Workspace with Gemini” as the paid upgrade that turns Workspace into an AI collaborator, adding things like generative assistance in

Docs/Sheets/Slides/Meet plus the new agentic workflows.

On the broader side, Gemini Enterprise on Google Cloud exposes agents as a platform across internal data and apps, but Workspace Studio is the opinionated, no‑code layer specifically for automating work in Workspace itself. As with other Workspace features, availability and pricing can vary by region and edition, and admins can control whether Gemini/agents are enabled for particular users or groups.

What you can actually do with them:

Common patterns Google is highlighting: automatic email triage, meeting and project update summaries, legal or customer-support document triage, approvals routing, and “concierge” style agents that coordinate travel or internal requests across multiple apps. Early customers have already executed tens of millions of tasks in a month with these agents, ranging from simple reminders to complex multi-step workflows that previously required manual coordination or custom scripting.

If you tell me what kind of workflows you’re running today in Google Workspace (e.g., lead intake, client onboarding, support routing, affiliate payouts), a concrete map of “here’s an agent you could build and how it would interact with your Workspace accounts” would be straightforward to sketch out.

Did you try it yet?

https://support.google.com/a/answer/13623623

https://studio.workspace.google.com/

2 comments