Skip to content
This FREE AI Tool Runs 100 Agents at the SAME TIME! artwork

This FREE AI Tool Runs 100 Agents at the SAME TIME!

AI News Today | Julian Goldie Podcast

October 16, 2026

This FREE AI Tool Runs 100 Agents at the SAME TIME!

Speakers Julian Goldie

TopicsMarketingBusiness

Julian Goldie (0:00)

Superset AI runs 100 agents at once. Superset AI lets you run Cloud Code, Codex, Hermes, and more than a dozen other AI agents at the same time in one app for free. And each agent gets its own desk. They can't mess up each other's work. You can check on all of them from your phone. Most people still run like one agent at a time. You give it a job, your wait finishes, then you give it a next job. Superset basically removes that waiting process so you can hand out like five jobs at once, go to the gym, come back to five finished pieces of work ready for you to check. And I call this the Many Desks Engine. It's a simple way to see how Superset works and why it changes how much work your agents can do in one day. By the end of this video, you'll understand the Many Desks Engine, the three moves inside it, and how to get access to Superset. So let's run through the numbers quickly first. Superset has over 15,200 stars on GitHub. It's open source. And the team have actually shipped 248 versions so far. The newest one, version 1.38, just dropped. And they ship pretty much every single day. Now, bear in mind, running one agent is super easy. Running five agents in the same folder is usually where everything breaks. So let's look at why it breaks, because once you see the problem, Superset basically makes a lot of sense. If you picture one kitchen with five chefs, one, for example, on the board, one on the stove, one on the pan, every chef is told to cook a different meal. What happens? Well, chef one basically puts down an onion, chef two moves it, chef three uses the pan that chef four is about to grab, and nobody's meal comes out right. That's what happens when you point five AR agents at one project. They all write to the same files. Agent one changes the page, agent two changes the same page a second later, and wipes out agent one's work. You end up with a mess, and you spend longer fixing it than you would have spent doing the job yourself. So most people do the same thing, which is just like one agent, one job, one folder, and then wait. And that waiting costs you a lot of time, of course. Kind of annoying. And also a lot of people don't like, for example, the remote option inside Claude. It's not that easy to manage from your phone, and that's why SuperSet fixes it, right? So the key to focus on here is the work tree. A work tree is a separate copy of a project that an agent can work on by itself. You want to think of it like this. Instead of five chefs in one kitchen, each chef gets their own kitchen, same recipes, same ingredients, but their own stove and their own pan. And you taste each meal and pick what goes on the menu. That's a work tree. So each agent gets its own copy of the files in its own branch, which is just a separate line of changes that nobody overwrites anybody.

And work trees have been around for years. The hard part was always using them. You had to type special commands, keep track of which copy was which, and open a new window for every agent. SuperSet basically does all of that for you with one click. Now, what actually makes SuperSet different from just opening up Claude Code inside a window, there are basically three things. The first is that it doesn't really care which agent you use.

So you can use this with Claude Code, Codex, Gemini, it works with any single AI agent, Hermes, Antigravity, whatever you want. And you can add your own custom agents with his own icon, and just works like the built-in ones. And that matters for anyone running more than one agent. For example, in my agentic operating system, Claude Code, Hermes, OpenClaw, it will run side by side because each one is good at different things. And SuperSet kind of works in the same idea. You can give Claude a big-built job, give Codex a fix, and give Hermes something else, all in the same window, so your workspaces, your branches, and your review flow stays the same no matter which agent you pick. The second thing is that your agents keep going when your laptop goes to sleep. So you can add any computer as a host. It could be a Mac Mini in the corner, it could be a big cloud machine, your workspaces run there. You could close your laptop, and the work carries on. And there's also an iPhone app. So you can start a job from your phone, and then check on it from anywhere. And to be fair about it, the phone app needs SuperSet Pro, iOS 26 or newer, and a computer with remote access turned on. The third thing is that you can create automation. So you can put, for example, jobs on the schedule, and the examples on the site, things like a daily check of new issues at 9am, a weekly update, monthly reviews. The agent runs on its own, and leaves the finish work waiting for you to approve. Now you might be thinking, okay, well, what's the difference here? Well, basically, with parallel sessions, the old way was one agent, one job, you watching. New ways, lots of agents, lots of jobs, each in its own space, some on a timer, and you checking the results wherever you are. Now, there's also a feature called pages. So an agent can publish a report to a link. Your team leaves comments pinned right on the report. An agent watching that page can fix it at the same link. So let's say, for example, you could have an agent write a monthly SEO report for a client. Your team leaves comments on it, and the agent updates you without sending any files back and forth. Split launch review. That's basically how it works. There's also a Linux version as well. So it runs on Mac, Apple Silicon, Intel, both work. And there's a Linux version, and Windows is not available yet, but I think it's coming. You can also bring your own agent subscription. So if you already pay for Claw to Codex, you just use those. SuperSet doesn't sit in the middle of your model course either. So you can also just plug in OpenRouter or a few other model providers too. By the way, if you want to learn how to use SuperSet in more detail, how to use Clawd Code, Hermes, all your agents, et cetera, then feel free to check out the AI Profit Boarding. You actually get the Agent OS system where you can plug in all your agents, Clawd, Hermes, OpenClaw in one place. Plus, you get my own tools for video, SEO agents, AI avatars already built in. You get the Zipfile video tutorial, the 30-day roadmap for setting up, and daily updates as we improve it, which we do daily. And also, you get four coaching course every week where you can bring your own agent setup, ask questions live about it, about how to run more of your work in parallel too. It also helps over 3,300 members inside the AR Profit Boardroom. Link in the comments description, or just go to the arprofitboardroom.com if you want to get access.

2 more minutes of transcript below

Thousands of transcripts fetched by people building searchable podcast archives

Fetch the whole transcript

The demo key returns a sample episode in full, no card needed:

request
curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

request
curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000796719406