← Back to search
NEW Antigravity 2.0 + Agent OS is INSANE!
AI News Today | Julian Goldie Podcast · 2026-05-20 · 19 min
Show full episode description
Hermes + AntiGravity Masterclass: Build an Agent Operating System (Context, Models, Migrations & Automations)Julian shares his setup for combining Hermes and AntiGravity so AI agents understand personal context and improve over time, then answers community questions about AntiGravity 2’s changes (removal of the built-in IDE/terminal), agentic workflows, and whether n8n is still worth using versus faster Hermes automations. He covers handling long-document context limits with model choice (including Claude), compacting, and chunking; recommends building a personal context engine in Obsidian over NotebookLM for long-term memory; discusses open-source/local model options via Ollama and LM Studio (including Hermes-focused and quantized models, plus Gemma as a lightweight choice); and explains migrating from OpenClaw to Hermes. He also outlines agency use cases like deploying SEO content, building/ranking sites, Twitter growth, and AI avatar marketing, and invites viewers to the AI Profit Boardroom for training, guides, coaching calls, and testimonials.
✨ Episode Outline — click any point to jump to it in the episode
Problem solved
How to make
Google Antigravity 2.0 far more powerful by plugging it into an agent operating system with persistent memory.
Benefits
- Orchestrate multiple autonomous agents in parallel
- Persistent memory so agents never start from zero
- One command center instead of 15 tabs
- Gemini 3.5 Flash for faster, better agentic builds
- Scheduled tasks, subagents and MCP customization
Use cases
- Built three websites in one day with Antigravity, including an AI Profit Boardroom landing page, in single prompts
- Memory system stores 1,553 memories about the user via Omi and Obsidian knowledge graph
- Tested from a list of 100 prompts to build HTML landing pages on demand
- Antigravity 2.0 runs Gemini 3.5 Flash for better images and agentic coding
- Claims the setup makes agents '10 times more powerful'
KPIs / results
- Three websites built in one day
- 1,553 stored memories about the user
- 100-prompt test library; claimed 10x more powerful
Tools / build
- Antigravity CLI inside Agent OS mission control
- Gemini 3.5 Flash plugged into Hermes via API
- Omi memory device + Obsidian knowledge graph
- Kanban board for autonomous agent teams
- Studio for image, video and speech generation
📑 Chapters — tap a time to jump there
00:00
Hermes Meets Antigravity
- Hermes meets Antigravity 2.0 in one command center
00:44
Antigravity 2 Changes
- Antigravity 2.0 changes: new UI, subagents, scheduled tasks
01:24
n8n Workflows and Training
- n8n workflows and agent training
01:39
Boosting Agent Context
- Boosting agent context with memory
03:25
Budget Friendly Setup
- Budget-friendly setup options
05:10
Beginner Automation Advice
- Beginner automation advice
08:11
Best Local Models
- Best local models discussion
12:05
Agency Stack and Processes
- Agency stack and processes
13:39
Join the AI Profit Boardroom
- Join the AI Profit Boardroom
Google Antigravity 2.0 just dropped in, your AI agents just became 10 times more powerful. This is Google's most advanced agent platform ever built. And today I'm going to show you how to plug it into a full command center. So you have an entire team of AI agents building websites, creating content, automating your work all at the same time. Now here's what makes this different. Most people open up Antigravity, they type it from, they get average results because they're missing this one thing. And I'm going to show you exactly what it is today. And when you set this up the right way, your agents know who you are, they know your business and they never start from zero again. So every single day they get smarter, every single output gets better. And every single day your AI operation improves over time. So by the end of this video, you're going to have Antigravity 2.0 running inside a full agent operating system, more powerful, more organized, and more automated than anything you've ever seen before. Stick with me to the end because there's one upgrade I'm going to show you that makes everything click. Today we have a brand new update from Google, which is Google Antigravity 2.0. So Google Antigravity 2.0 is a dedicated platform to work with AI agents. You can orchestrate multiple autonomous AI agents working in parallel across independent tasks. Basically what that means is you can use this, it looks very similar to Codex now, they've totally changed the UI. Basically this allows you to create whatever you want. So you can build out websites, you can build out teams of agents, you can have it organized for websites for you. So you can see, for example, I built out this full website right here using Antigravity. And with Antigravity 2.0, it's also got Gemini 3.5 Flash plugged in, which is super powerful stuff, right? Now also on top of that, the new updates include dynamic subagents. So subagents are defined basically as like a team of agents that work together, right? Then you've also got scheduled tasks. So you'll see over here, you have scheduled tasks, and you can give it new tasks to schedule and automate routine checks, which is pretty cool. You have artifacts. So these are deliverables from the agent to communicate its progress with you. And you can also extend customization, which is pretty powerful. Top of that, you get projects. So you can actually group your projects together, which you can see over here. And it has live voice transcription. So you can speak your prompts, and then it will start generating the prompts from your voice. And additionally, you can add more customizations inside here, because you can now set up MCPs inside Antigravity to build whatever you want and automate whatever you want, which is pretty powerful. Now, the way that I'm using this to get the most out of it is I've actually plugged it into my agent operating system, which you can see right here. And the reason for that is because it's way easier to manage. I can see all of my history, and I can see everything I've built inside one beautiful workspace here, right? So we actually built three things with Antigravity today. Now, the way that I'm doing that is I have this agent operating system, and it's connected to Antigravity CLI. You're not sure what that is. So basically, Google have a Gemini CLI, and they're actually changing that to Antigravity CLI, right? And so what this means essentially is you can speak to Antigravity in the same way you would speak to Claude Code or to any sort of interface. Now, if you are running Antigravity inside the terminal, it just wouldn't look that nice. Let me show you an example of that. So if we open up our terminal here, we can operate Antigravity inside a terminal using the Antigravity CLI, right? And you can download it here. But the problem with this is that it's not that nice to use inside the terminal, and also doesn't interact with the other AI agents, right? And so how do you get a better web user interface? How do you get something that looks a lot nicer? Here's what I'm doing. So what I've actually built out here is Antigravity CLI inside an agent operating system. It's Agent OS, right? And so what this means is that we can use Antigravity inside our mission control dashboard along with Hermes, along with Gemini, along with, for example, Claude and all other AI agents. Now, why is that important? It's important because now we can have all of our agents communicating together, working together, and just building whatever they want. And also, this is attached to my memory system. So what that means essentially is it has full context of me, who I am, what I do, et cetera. And it's all linked to my Obsidian dashboard right here, right? And so every memory about me, every person in my life, everything that I work on, et cetera, is all inside this setup. And then I can plug that into Antigravity. I can plug it into Gemini CLI. I can plug it into Hermes, et cetera, and build and automate whatever I want. Super powerful stuff. And so this is what I would say is the best way to handle your AI agents inside one place. So you can see, for example, if we have a look, I can actually see all the websites that I've built previously, all the content I've created. And I've even got, for example, a studio for generating images, video, and speech. This has Notebook LM put in as well. So I can combine Antigravity with Notebook LM to pull in, for example, like podcasts or videos or infographics that I've created with AI. Then we have a Kanban board where I can actually assign teams of agents. And we have all of our AI agents working together, right? Really powerful stuff. You can see an example of a website that we actually created. Super clean, looks nice. And we just built this out with Antigravity in one prompt, right? How do we do that? Basically, what we said was build me a landing page for my coaching program here. And then it went off and just built out this beautiful website, as you can see. And the cool thing about this is it has Gemini 3.5 Flash plugged into Antigravity, which means that you get better images, you get faster outputs, it's better at coding, it's better at agentic tasks. Google have really designed Gemini 3.5 Flash to be an agentic model. And so with this new version of Antigravity, it's more powerful than ever before, particularly when you plug it into an agent operating system. Marco says, can you use your Gemini inside Hermes? So you would have to, I think as far as I'm aware, you would have to plug in your API, which is what I've done here, right? So I've plugged in Gemini 3.5 Flash directly into Hermes Agent. And that's how we can use Hermes Agent with Gemini, right? Whereas for example, if you, I don't think you can use OAuth with a subscription, you'd have to plug in API. So let's talk about the operating system, how we did it, et cetera. So basically what you can do is you can get Antigravity CLI, and then you can connect this to your agent operating system. Now, if you just want to get the full setup for me, et cetera, we've got a full guide with prompts, tutorials, and a video guide over here. So if you want to get everything that's inside the iProfit board, and we have the prompts, the video tutorial, and also the files attached in a zip file. So you can just implement that and get that agent operating system, or you could get Antigravity to actually build it for you, but it might take a bit of time, a lot of tweaking, et cetera. So what we can do with Antigravity now is we can just give it any tasks we want to. We've actually got a list of 100 prompts that I could test over here. Let's have a look at this. So we could say, okay, build a beautiful HTML landing page for this, right? And then we plug in the details. So if I go inside the chat here and I'm saying, okay, build a beautiful HTML landing page for the iProfit boardroom. And then I'm just going to give some context on that, which we've got over here. It will just go off and do that, right? And that's basically how you can use this so easily. That's how we built out this website. We actually created some other stuff. So if we have a look over here, you can see that we have three different websites built out, right? So here's another one. Let's open this up. And the cool thing about Antigravity as well is it can generate images, right? So it can generate nice images for you too. And here's the full website created with Antigravity 2.0, right? Super easy to do. And then here's another example of a website we've built out. This is for the iProfit boardroom. If we open this up, this is a landing page that we built with Antigravity 2.0. You can see it's just nicely designed. It's clean. He literally created the website, designed it, created all the images, put it all together. And now it's looking awesome, as you can see. So super easy and simple. Now we can also plug in Gemini into our CLI 2.0. So we've got Antigravity and Gemini in terms of maybe you want one for just general coding stuff. And Antigravity is more for like agentic tasks. But we also have Hermes here, as you can see, we have OpenClaw, we have Clawed, whatever we want to use. And so whatever you want to build with this, you can check out your files, you can use the chat, et cetera. And it's super powerful and easy. Now you might also wonder, okay, why have we created an agent OS system with Antigravity, right? Here's the way that I look at it. So an agent operating system is really just a command center for your AI agents, right? They all work together. And so you can have them all integrated into one place, super simple and easy to set up, and it works nicely. And also you can just integrate something. So some of the tasks I typically work on day to day, I've actually set up as like separate tools that integrate with Antigravity and with Gemini over here, right? So for example, what you can see here is that we have our memory plugged in. And so it has loads of memories on me, who I am, what I work on, that's pulled in from Obsidian, right? And so Obsidian and my AI agents organize a knowledge graph about me, make it look beautiful, and then turn that into an amazing system where we can give memories and give context to our AI agents, right? It's literally just taking notes on me all day. Now, if you're wondering how we do that, basically we have a memory system with Omi, which is taking in information about me, what I do all day, what I'm working on. So you can see here is taking notes on like how I'm working on Antigravity today with Gemini 3.5 Flash. And it's taking that information and then using that to take notes on me and organize information about me. Then we export it to Obsidian and with Omi taking notes on me and then plugging it into Obsidian, we have a really powerful base of memories here. We have 1,553 memories about me, who I am, what I work on, et cetera. Then that goes into a beautiful knowledge graph like you can see. So it's super organized. Everything links together. It has a lot of information about everything in my life. And finally, that plugs into our agent operating system, right? Really powerful stuff. So let's talk about why Antigravity is 10 times more powerful inside the agent OS, right? So most people are using Antigravity wrong because they open it up, they type a prompt, they get an answer, and they wonder why the results feel average. This guide really shows you a different way to use it. It's one that stacks Antigravity inside a full agent operating system, one that plugs in a memory system using Omi and Obsidian, one that makes every agent you run permanently smarter over time. So by the end of this guide, you understand exactly why this setup is more powerful than anything else out there. So let's talk about this first, right? What is Antigravity? Antigravity is Google's first agent-first deployment and development platform, right? This is Antigravity. And this is just like the basic program, not the agent operating system, right? And you want to think of it as a launch pad for your AI agents. So instead of just chatting with AI, Antigravity lets you deploy AI, right? It has multiple agents working at the same time, all on your goals, all supervise you. So when coupled with the updated Antigravity harness, it becomes a powerful engine but just having your sub-agents work on a problem together. Now you might be saying, okay, but what is the agent operating system then? So this is a command center for every AI agent you run. Think of it like your phone's home screen, but instead of apps agents, right? You can see what they're working on. You can see what they've already built. You can see what files have created, right? We can see that, for example, for Antigravity over here, or for example, for Hermes with Grok 4.3, we can see that here. For any agentic autonomous teams that were built, we can see it on the Kanban board and see what tasks have been done. So you can see what they're working on, what they've built, what they've created. You can launch new tasks with a single prompt and you can manage everything from one place. So instead of juggling 10 different tools across 15 tabs, right? So for example, going over to Antigravity here and then going over to Terminal to use Hermes and then switching to another tab to use Claude, that would be super messy, right? With this, you have one dashboard, one command center, everything organized, everything tracked and everything running together, right? Now at this point, you might be wondering, okay, what is OMI? So OMI is an open source AI memory device and app, right? It sits on your desk or you can actually wear it as well. They have wearables too. And essentially it listens to your day, right? It captures everything that you say, you learn, you think about it, transcribes it, it turns it into structured memories so that every meeting, every idea, every conversation is all stored or captured and organized, right? So you can see that inside the memories here of OMI, which is pretty cool. Now you might be saying, okay, what is Obsidian? So Obsidian is a free note-taking app like this, and it's stored as plain files, right? So you just have a folder of your most important thoughts. And so when you connect OMI to Obsidian, all those captured memories flow into a single vault, a living, breathing second brain, right? That you own forever. And so what I call this is the Goldie Gravity Grid, which is a new paradigm for running AI-powered operations in 2026. So most people are using AI as a tool. The Goldie Gravity Grid turns it into an orbit, right? With every layer pulling the others higher and improving them over time. So for example, if you look at like how people, how 99% of people use AI right now, if they're going to go into anti-gravity, they would see it like a hammer, right? So you pick it up, you use it, you put it down. But that mental model doesn't really get you great results, right? The Goldie Gravity Grid is a completely different paradigm because it's not about using AI, it's about building an AI orbit around your work, right? Each layer pulling the others higher, each layer making the next one more powerful. And unlike a hammer, it gets stronger the longer you run it. So let's talk about the gravity grid. So the first layer is the capture layer, which is OMI, right? And this is where your intelligence gets stored. So every conversation, every meeting, every idea gets captured by OMI passively, right? And without you doing anything extra, it's just taking notes on your day. And this is the raw material your entire AI orbit is built on. There's no, there's no, if you don't have any way to capture your thoughts, then you don't have any context. If you don't have any context, you get generic outputs for your AI agents, including anti-gravity. And so with generic outputs, you get wasted time. OMI fixes that at the root. Then the second layer is the vault layer, which is obsidian. So with obsidian, this is your second brain. It's structured, it's searchable, it's permanent. So your business context lives here, your clients live here, your goals live here, your brand voice lives here. And the vault gets smarter every single week. And it's all on your computer. So it's private, secure, it's yours. And then you've got the intelligence layer, which is anti-gravity. So anti-gravity on its own is okay. Anti-gravity with the full OMI and obsidian stack is 10 times better, right? So when you bring in anti-gravity, you can point anti-gravity at your obsidian vault. And suddenly your agents know who you are, they know your business, they know your goals, they know your clients, they know your voice, right? And this is what turns anti-gravity from a generic AI platform into a personalized AI command engine. And then you have the command layer, which is agent OS, right? Agent operating system. And so this is where you can orchestrate a thing. It sits on top of anti-gravity. It gives you the dashboard to deploy, monitor, and manage every agent. You don't just run agents, you run fleets of them, right? Each one pulling from your obsidian vault, each one powered by anti-gravity, each one doing a different job simultaneously. So you could use this for SEO, for research, for outreach all at once, right? If we said, okay, for OpenClaw, do content research and SEO, it would just go off and do that for us, right? And that's how it's building all these websites that are super personalized to us. And then finally you have the loop, the layer five, right? So every output your agent produces gets fed back into the system, right? So more outputs equals more context in the vault, which is what we've got over here. More context equals smarter agents, especially inside the memory vault. Smarter agents equals better outputs, and better outputs equals more improvements. So this is a flywheel that most people never build. Most people just start over with every prompt, but the goldy gravity grid means you never start from zero again, right? And so the thing that I would say is you don't build a business with one great prompt. You build it with a system that gets smarter every single day, and that's the biggest difference. Now you might say, okay, why would I use, for example, anti-gravity inside agent OS instead of running it inside this system here? The core problem with standalone anti-gravity is like when you run it on its own, it's got no memory. So every session starts cold. Anti-gravity doesn't know who you are, doesn't know your business, doesn't know your last conversation. And you spend 20% of every prompt just re-explaining yourself. Number two is there's no reorganization here, right? When you look at this, you can't see your outputs. It's pretty hard to find your website and you can't really view, you don't have that IDE view that you have anymore, that you used to have, right? Whereas inside anti-gravity, you can see all your files here. You can easily go to them, you can open them up, you can see everything that was created. We can quickly navigate to the whole project and we can quickly preview everything that was created, right? Big difference right there. And also it doesn't improve, right? So with anti-gravity, if you use it standalone, you use it, you get a result, but that's it. The next session is the same as the first session. Nothing grows, nothing builds, nothing stacks. So what agent operating system does is it wraps around anti-gravity or moves all three walls. So adds a memory via Omin Obsidian adds organization via the dashboard and workspace structure, and it adds improvements, the loop via stored workflows, reusable prompts and growing context. So running anti-gravity inside AgentOS is like upgrading from a bicycle to a motorbike. The engine is the same AI, but the vehicle around it changes everything. So that is the system. If you have any questions, if you're watching live, feel free to ask. But the other thing I would say here is like, a lot of people are going to look at this and think, I don't know where to start or whatever. I would say your results are a direct reflection of the beliefs you hold about what's possible, right? So if you change the belief, you can change the output. Let me show you an example. So some people say, I'm not technical enough to set this up. Here's the cool thing. You can actually get the full setup from the AR Profit Boredom community here with all the files, the prompts, everything that you need, right? We've got a full tutorial on how to set up. Literally, I set up pretty quickly and ever since, I've never looked back, right? So if you can copy and paste the file path, you can run the system, right? The barrier to entry here is not technical knowledge. It's following a step-by-step process. Other people say AI tools are all the same. There's no advantage to setting this up differently. I would say the architecture around this is actually more powerful than the AI itself, right? If you have a look at this, it's pretty interesting. So if you had a Formula One car engine, aka like Gemini 3.5 Flash, in a normal car, it would still lose a race. But the structure, the setup, the system, these are what win. So two people can use the same AR model. One uses it just like a chat and one uses it inside the Goldie Gravity Grid. And the results are not comparable, right? It's the same engine, but a completely different vehicle. For example, like yesterday, we built out a whole new website in one click using the Kanban theme section. We can create videos and podcasts directly inside the chat here. We can generate images and videos and text-to-speech, right? You can't really do that inside Antigravity in the same way, right? It's just not, it's not going to be organized in the same way. And so it's the same engine again, but a completely different vehicle. And other people will say, I don't have time to set up a whole new system. I would say you don't have time not to, right? Because if you can set this up day one, then the time saved starts today, right? We've also got a 30-day roadmap on how to implement this, which you can see down here, right? So we have a full step-by-step operating procedure and a 30-day roadmap on how to implement it. But yeah, the more you use it, the more powerful it gets. So that's basically it. If you want the full guide, the setup, et cetera, I'll put it inside the AR Profit Board, and you can actually see it here. It's got a video tutorial. If you want the full agent operating system with Hermes and Claude and everything, we've got over here, and we have all the prompts, a guide for implementation, and the files that you can use to set this up. And that just makes Antigravity 10 times more powerful, especially with the new updates.