Google's agent builder is here

Edition 142 | December 8, 2025

AI agent: I scheduled your meeting, responded to 12 emails, cleaned your inbox.
Me: ‘Can you open Calculator?’
AI agent: ‘Look, I’m a language model, not a miracle worker.

Thankfully, we do have models that can use calculators natively.

Welcome back to Building AI Agents, your biweekly guide to everything new in the field of agentic AI!

In today’s issue…

• Google launches Workspace Studio

• 5-step guide to building an AI lead generation agent

• First documented AI-led cyberattack (GTG-1002)

• LangChain opens agent builder in open beta

• How to build a no-code SEO AI agent

…and more

🔍 SPOTLIGHT

Last week, Google announced the general availability of Workspace Studio, their platform that lets anyone build AI agents directly inside Gmail, Docs, Sheets, Drive, and Calendar using plain English. Powered by Gemini 3, users describe what they want an agent to do ("If an email contains a question for me, label it 'To respond' and ping me in Chat") and the agent gets built instantly, no code required.

For agent builders, this is a seismic shift in what "accessible" means.

Until now, the agent-building landscape has been split. Developers use frameworks like LangChain, CrewAI, or OpenAI's Agents SDK to wire up tool calls, manage memory, and orchestrate multi-step workflows. Non-technical users get rigid if-then automation tools like Zapier that can't actually reason. Workspace Studio bridges that gap by giving everyday users agents that can understand context, adapt on the fly, and execute complex workflows—capabilities that previously required engineering effort. Early testers have already run over 20 million agentic tasks in 30 days.

What makes this different from other no-code agent builders is the native integration. The hardest part of building useful agents isn't the reasoning, it's connecting them to where work actually happens. Workspace Studio agents can read emails, extract data from attachments, update spreadsheets, check calendars, and trigger follow-ups across the entire Google ecosystem without a single API key or OAuth flow. For anyone who's spent hours debugging broken integrations, that's the real unlock.

The business model matters too. OpenAI and Anthropic charge $20/month for their pro tiers that give you a better chatbot. Google is bundling full agent-building capabilities for free into Workspace subscriptions that companies already pay for. As the All-In podcast discussed this past Friday, Google is using its cash reserves to subsidize AI adoption, and agents are now part of that bet.

For those building in the agent space, this changes the calculus. Simple workflow agents (email triage, meeting summaries, document routing) are now table stakes. The opportunity shifts toward domain-specific agents, complex multi-system orchestration, and use cases that extend beyond Google's walled garden. Workspace Studio supports external connectors to Asana, Jira, Salesforce, and custom APIs—but Google just set the baseline for what users will expect out of the box.

The agent wars won't be won by the best model alone. They'll be won by whoever gets agents into daily workflows first. Google just embedded them into a productivity suite used by over 3 billion people.

Always keep learning and building!

—AP

P.S. Roll out of Google Workspace Studio has just started. We will be launching courses & guides to help you navigate Workspace Studio, stay tuned!