01What the Hermes Subagents Update Is and How It Works

The Hermes subagents update lets you run multiple AI agents at the same time in the background. Instead of waiting for one task to finish before starting another, you can now delegate work to a hive of sub-agents that work in parallel.

This works through a new delegate tool. When you give Hermes a task, it can create multiple sub-agents at once. Each sub-agent gets its own thread, its own goal, and its own context.

They all work simultaneously, and results fly back to your main chat as they complete.

In this episode, Julian Goldie explains how this update changes the way you work with AI agents. You no longer have to watch your chat freeze while a task runs. Instead, you hand off work to sub-agents and keep going with other things.

02Main Features of the Hermes Subagents Update

The update brings several key features that make Hermes more powerful for real-world tasks.

  • Asynchronous sub-agents: Sub-agents run in the background without blocking your chat.
  • Delegate tool: Hermes uses this new tool to create and dispatch multiple agents at once.
  • Parallel execution: You can run five or more sub-agents at the same time, each handling a separate task.
  • Fast dispatch: With background=true, Hermes hands control back to you in just 2 milliseconds.
  • Self-contained results: Each worker finishes independently and re-enters your chat as a new turn.
  • Plain English control: You don't need coding skills; you can speak to Hermes in normal language.

These features let you automate complex workflows like writing multiple blog posts or running an entire SEO funnel without babysitting each step.

03How to Set Up the Hermes Subagents Update

Getting started with the subagents update is quick. Julian shows two easy ways to update Hermes.

  1. Open your terminal and type 'Hermes update'.
  2. Or go to your Hermes dashboard and click the update button.
  3. After updating, you can start using the delegate tool in your tasks.
  4. If you want to manage agent profiles, you can do that in the dashboard as well.

Julian notes that the update was released just this morning, and he already set it up successfully. He says, 'If we can do it, you can do it too.'

You can also let Hermes set up sub-agents for you, or you can plug in Claude if you prefer. In his agent operating system, Julian uses multiple different agent profiles, and when he gives a task for sub-agents, Hermes builds them in the background automatically.

04How to Use Sub-Agents with the Delegate Tool

Using sub-agents is simple. Give Hermes a goal, and it handles the rest.

For example, you could say, 'Run an SEO funnel for how to build an AI agent.' Hermes will use the delegate tool to create multiple workers that work together to create something awesome.

You can give Hermes one keyword and have a hive of AI agents working in parallel on five different articles. Then, a deploy worker can take the content and publish it to your website.

Here is a step-by-step example from the episode:

  1. Give Hermes a task like 'delegate a keyword research task to five sub-agents working in the background.'
  2. Hermes dispatches the sub-agents using background=true.
  3. Each sub-agent runs its own thread with its own goal and context.
  4. Workers finish at different times, and each result re-enters your chat as a new turn.
  5. You can keep chatting or start the next keyword while the agents work.

You can also use different APIs for sub-agents. For menial or admin tasks, you can use a cheaper API, a free one, or even a local one, and it still works.

05Results and Comparisons Shown in the Episode

Julian demonstrates the power of the update with a real example. He set up five sub-agents to create five different articles for his website. They all worked in parallel and deployed the content automatically.

He compares this to the old way of doing things. In the old way, you would write one article, wait for the next one, and check each one. Each task blocked the next one, and you had to sit and watch as the chat was frozen.

With the new update, you give one command, five workers start at the same time, they all write articles, you walk away, and everything comes back on its own. The result is a whole funnel built while you do something else.

Julian also notes that each worker gets a tight goal in its own context, and every result flies back for you to check. You are the lead; you set the jobs and then walk away.

06Tips, Limits, and Mistakes to Avoid

Here are some important tips and warnings from the episode:

  • Don't watch every step: The whole point is to let sub-agents work on their own. Trust the process.
  • Give tight goals: Each worker needs a clear, specific goal to stay on track.
  • Check results as they come back: Results re-enter your chat, so you can review them when they arrive.
  • Don't worry about chaos: Running agents in parallel might sound chaotic, but it works like a background terminal job.
  • You don't need to be a coder: Speak to Hermes in plain English, and it will build what you need.
  • Update regularly: Make sure you run 'Hermes update' or click the update button to get the latest features.

One common mistake is trying to micromanage every sub-agent. Instead, let them work independently and collect the results as they come in.

07Who Should Use the Hermes Subagents Update

This update is for anyone who uses Hermes for content creation, SEO, or any task that involves multiple steps. If you are a business owner who needs blog posts, you can use sub-agents to write and deploy articles automatically.

It is especially useful for people who want to scale their output without hiring a team. With sub-agents, you can run multiple projects at once, like an SEO funnel, and get results while you focus on other priorities.

Julian emphasizes that you don't need coding skills. In his case, he and many others built their own agent operating systems using this setup, and they are not coders.

The update also integrates with the broader AirProfit One system, which turns Hermes, Grok, Build, Claude, GLM, and Kimmy into one system with shared memory and a single dashboard. Every new feature, like asynchronous sub-agents, makes the whole system more powerful automatically.