- Practicaly AI
- Posts
- The Agent Era Is Here: Google, Microsoft, OpenAI Move In
The Agent Era Is Here: Google, Microsoft, OpenAI Move In
Today in AI: OpenAI launches Workspace Agents, Microsoft makes Copilot agentic, and Google connects Workspace with AI.
👋 Hello hello,
We’re officially past the “AI assistant” phase. What we’re seeing now is something more serious — AI that can operate, coordinate, and do the work across systems.
Where: inside your tools, your docs, your spreadsheets; the places where work already lives. Today’s updates are about control layers, shared context, and systems that don’t need constant babysitting.
💬 Quick note: We’re building something to help teams actually get good at AI (not just use it). → Get early access here
🔥🔥🔥 Three Highly Curated AI Stories
OpenAI is rolling out workspace agents in ChatGPT, essentially Codex-powered agents for teams.
These are shared agents that can handle long-running, multi-step tasks across tools and teams. You define the job once, and the agent keeps things moving — pulling context from docs, emails, chats, and even external systems.
They can qualify leads, route feedback, review requests, pull reports, or research vendors. They can also take actions — like updating Linear tickets, sending messages in Slack, or creating documents. The key shift is that these agents don’t stop when you close your laptop. They keep running in the background or on a schedule.
Satya Nadella announced it quietly, but the implications aren't quiet at all: Agent Mode is now the default experience inside Copilot for Word, Excel, and PowerPoint.
The shift matters because of where it lives. When an agent has access to how your data is actually structured — relationships in spreadsheets, logic in models, narrative in documents, it can reason across all of it at once, not just answer questions about it.
You stop prompting for small outputs. You start reshaping entire workflows. That's either very exciting or slightly terrifying, depending on how much you like controlling your own Excel files.
But if you’re only getting warmed up with Copilot, here’s where I would first start.
Google is introducing Workspace Intelligence — a unified AI layer across Workspace powered by Gemini.
The goal here is to remove context silos. Instead of Gmail, Docs, Sheets, etc. working in isolation, this layer connects them so AI can understand and act across your entire workflow. Think of it as one system that sees everything and helps you move faster without switching tools.
This sounds very similar to Microsoft’s Work IQ direction — both are trying to build a shared intelligence layer across productivity tools.
The move makes sense. Google has the cloud, the models, and the enterprise contracts. The question is whether anyone outside a Fortune 500 will actually use it.
🔥🔥 AI Signals You Shouldn’t Miss
If you’ve been hearing terms like “reactive agents,” “autonomous agents,” or “multi-agent systems” and just nodding along — this is for you.
This glossary breaks down the different types of AI agents in plain English, with clear distinctions and use cases. It’s especially useful if you’re trying to figure out what kind of agent you actually need (instead of building blindly).
Google just introduced a full-fledged platform for building and managing AI agents with Google Cloud. This is essentially the next evolution of Vertex AI — but now focused heavily on agents instead of just models. The platform brings together model selection, agent building, integrations, and governance into one system. You also get access to 200+ models through Model Garden, including Gemini 3.1 Pro, Gemini 3.1 Flash Image, Lyria 3, and Gemma 4.
Google is positioning agents as something companies build and scale, not just experiment with. This is infrastructure — not a feature.
🔥 One AI Tool Worth Trying Today
🎨 DESIGN.md Directory
This tool is essentially a directory of machine-readable design systems — basically visual identities that AI agents can understand and follow.
Instead of manually prompting design styles every time, you give the agent a DESIGN.md file, and it works within those constraints.
What makes this useful is that it:
Works across tools that generate UI, websites, or visual assets
Lets teams standardize design outputs across AI workflows
Can be installed directly into your codebase using an npx command
Reduces the “randomness” in AI-generated design
Best for: teams building with AI (especially devs + designers) who want consistent outputs without rewriting prompts every time.
Bonus: If you want to understand how that works in real-time and how to use a design.md file for UI, this video from Google provides a quick demo:
💬 Quick poll: What’s one AI feature you tried recently that genuinely surprised you?
Did you learn something new? |
Until next time,
Team @PracticalyAI
Reply