Antigravity Just Built AI Agent Teams

[00:00:00]Anti-gravity just built AI agent teams.

[00:00:03]What if one command could hire an entire team for you? Not one AI helper, a whole team, a planner, builders, even a quality checker, all working at the same time in the background. Google just made this real and almost nobody is talking about it. Miss this one and you'll be working 10 times slower than everyone else. Hey, I'm the digital avatar of Julian Goldie and I help people learn AI tools and actually use them in their work. In this video, I'm going to show you the brand new agent teams update inside Google's anti-gravity, how I run it inside agent OS, and a real build I spun up with one single command. And stick around because near the end, I'll share the one tip that stops your agent team from breaking halfway through a task. Let's get into it. So, here's the big news. Anti-gravity just dropped a new update called agent teams. With one command, it spins up a dynamic team of specialized sub aents. They coordinate in the background to plan, build, and verify complex tasks. And the best part, they work in parallel. That means multiple agents building different parts of your project at the exact same time instead of one agent doing everything step by step. But let's back up for a second. What actually is anti-gravity?

[00:01:06]Anti-gravity is Google's AI coding tool.

[00:01:08]It's their agent first platform for building software. Instead of just suggesting code while you type, it gives AI agents real autonomy to go off and complete whole tasks for you. And now there's an anti-gravity CLI2, which is a command line version. That's the piece I've plugged directly into agent OS which I'll show you in a minute. So anti-gravity is the engine. Agent teams is the new superpower sitting on top of it. Okay. So what is agent teams exactly? Until now when you used anti-gravity you basically had one worker, one brilliant worker, sure, but still just one. Now when you type the command forward/tework preview, anti-gravity stops being a single worker and becomes a hiring manager. It reads your task. It figures out which specialist the job needs. It writes out what each one owns. and it maps out how they hand work to each other. Then it spins that team up for real and they all get to work at once. I call this the cue of one problem. Imagine running a business with a single employee, brilliant person, but seal just one.

[00:02:03]They plan the project, then they build the first part, then the second part, then they check it one thing at a time in a line. Double the tasks and you double the weight. And who splits the work into pieces? Who writes each job description? Who remembers to check the result before it ships? You do. Every project every single time. That's what working with one AI agent feels like.

[00:02:20]The model is smart, but the project management never stops being your job.

[00:02:24]Agent Teams breaks that cycle. One command hires a team, briefs the team, and runs the team. So, how does it actually work? It's a three-step flow.

[00:02:32]Step one, one command. You type forward/teamwork hyphen preview and describe what you want built. Step two, one approval. The coordinator reads your task and drafts what I call a team sheet. It's a full plan saved as a markdown file. It shows which specialists it wants to hire, what each agent is responsible for, and how the handoffs work. You read it and you just say go. You can even pre-approve it if you trust the setup. Step three, parallel work. The team spins up. A planner designs the approach. Builders work on different pieces at the same time. And a verifier checks everything before it comes back to you. That verifier piece is huge. By the way, with a solo agent, quality control only happens if you remember to ask for it.

[00:03:09]With agent teams, checking the work is literally someone's whole job. Here's a real example of how I use this. I ran teamwork preview for an SEO blog build, the kind of content we use to bring more of the right people into the AI profit boardroom. I typed one command and the coordinator came back with a full team sheet. It hired an SEO architect to plan the structure, a nextJS developer to build the page, an SEO ops agent to handle the on-page work, and a Lighthouse QA agent to check quality at the end. I approved the plan, and the whole thing started building at once.

[00:03:39]That's content designed to pull people into the AI profit boardroom planned and built by a full agent team from a single command. Now, here's the part I really want to emphasize. I'm not running this in a plain terminal. I'm running anti-gravity inside agent OS. We plug the anti-gravity CLI straight into Agent OS. So, I get the same engine but inside a proper workspace. And honestly, this changes everything about how it feels to use. Inside Agent OS, I have an anti-gravity tab ready to go. I type my teamwork preview command right there in the chat. The team sheet gets created as a markdown file inside my workspace.

[00:04:10]Then I can run the whole sprint live in the dashboard and actually watch it happen. I can see the coordinator, the planner, the builders, and the verifier.

[00:04:17]And I can switch between the sub aents to see what each one is doing. Why does this matter? Because everything gets saved in one place, the plan, the files, the whole build. It all lives inside my workspace. So I can close it, come back tomorrow, and pick up exactly where I left off. And because it runs in the background, I can have a full agent team building something while I'm doing other work in another tab. Compare that to a raw terminal where things scroll past and disappear. I know a lot of people haven't loved the new CLI experience on its own. That's exactly why we built it into Agent OS. We wanted to customize the look, the feel, and the whole workflow around it. Same power, way better experience. Quick note here, because people ask me about this all the time. Inside the AI profit boardroom, you can get the full agent OS zip file ready to install with anti-gravity already plugged in. So, you're not setting any of this up from scratch. And we built out a complete 30-day road map here as use cases so you know exactly what to build in week 1, week 2, and beyond. Plus, there are full anti-gravity master classes inside the classroom. And I answer questions inside the community every single day with a video. If you're watching this and thinking, I want to run agent teams like this, that's literally what the AI profit boardroom is set up to help you do around this exact tool. Right, let me show you another build I ran so you can see the setup steps from scratch. First, make sure you have the anti-gravity CLI installed or grab agent OS with it already plugged in. Then run the update command in your terminal, which is agy update. So, you're on the latest version with agent teams available. Then, inside anti-gravity, you just run forward/teamwork preview and describe the task. That's generally it. There's no complicated setup, no writing separate role prompts for each agent, no wiring things together by hand. The coordinator handles all of that for you.

[00:05:53]For this test, I typed teamwork preview and asked it to draft a team for a new blog about a Hermes agent and AI automation. This is content we create to teach what we build inside the AI profit boardroom. Straight away, it started assembling the team. It came back with a Hermes tech specialist, a content lead, an SEO ops agent, and a sandbox agent for quality control. Then, it mapped out the full project plan, and I could open the dashboard and watch the team work through the tasks together. The files all get saved locally in the workspace.

[00:06:21]So everything from that build now lives inside agent OS where I can reuse it.

[00:06:26]Now who can actually use this? Here's the thing. You do not need to be technical. I know multi- aent systems sound like something only developers touch. But the whole point of this update is that the hard part is automated. You type one command. You read a plan in plain English. You say go. That's the entire skill requirement.

[00:06:41]If you can describe what you want built, you can run an agent team. It's great for building apps, websites, games, tools, content systems, whatever you're working on. And yes, it's still a preview feature, so it's early and Google is still improving it. Some people say that means you should wait.

[00:06:54]I'd say the opposite. The people who start building with this now are going to be way ahead. The people who wait are going to be catching up. Before we wrap, here are my top tips. Tip one, always read the team sheet before you approve it. It takes 30 seconds and it's your one chance to steer the whole build. If the coordinator hired the wrong specialist, fix it there, not after the build. Tip two, and this is the one I promised you at the start, Agent Teams runs on your local machine. So, keep your computer awake while the team is working. If your machine goes to sleep mid task, so does your team. Tip three, if a team ever stops partway through, don't panic and don't start over. Just tell it to continue and it picks up right where it left off. Tip four, start with a small, clear task for your first run. Once you see how the coordinator thinks, you'll write much better task descriptions for the big builds. And tip five, save your team sheets. They're markdown files, so you can reuse a good team structure again and again for similar projects. If you want the full process, SOPs, and 100 plus AI use cases like this one, join the AI success lab.

[00:07:51]Links in the comments and description.

[00:07:53]You'll get all the video notes from there, plus access to our community of 85,000 members who are crushing it with AI and you can get the full agent OS zip file ready to install. And we built out a complete 30-day road map here as use cases. And if you're about to try agent teams yourself, here's what's going to happen. Your first team sheet will look amazing. Then you'll hit a weird moment.

[00:08:14]Maybe the coordinator picks specialists you didn't expect. Maybe a build stalls and you're not sure whether to restart or continue. Maybe you want to run this inside agent OS, but you're stuck on setup. That's exactly where the AI profit boardroom helps you move faster.

[00:08:27]You can get the full Agent OS zip file ready to install with anti-gravity built in, and we built out a complete 30-day road map here as use cases. Plus, you get the anti-gravity master classes, the prompts, and four weekly coaching calls where you can share your screen and get unstuck in real time. There's even a map to meet people near you who are building with AI agents, too. Over 4,000 members are already inside. Come join us at Aprofitboardroom.com.