00:00:00The developers of Claude Code just silently announced massive updates that almost no one
00:00:04is talking about.
00:00:05The biggest of which is the release of async background agents.
00:00:09What's strange is that you won't find these features mentioned anywhere on Anthropic's
00:00:13main site or even their blog.
00:00:15Now you might be wondering if these are actually useful, and honestly I wasn't sure at first
00:00:19either, but I've found an unexpected angle that's completely changed how I use them.
00:00:23So in this video, I'm going to break down exactly what that is, why it makes a lot of
00:00:27existing AI coding frameworks basically unnecessary, and then walk through several other Claude
00:00:32features.
00:00:33They announced these new features in the changelogs on Claude Code's GitHub repo.
00:00:37They mentioned that these agents and bash commands can run asynchronously and send messages to
00:00:42wake up the main agent.
00:00:43The idea here is to achieve true parallel working.
00:00:46It is quite simple to use them.
00:00:48We first ask Claude to handle a task that benefits from a sub-agent, and then the main agent creates
00:00:53a sub-agent for the task.
00:00:55When the agent has started running, we push it to the background using Ctrl+B and continue
00:00:59working with the main agent on other tasks.
00:01:02When the sub-agent has completed the task, it wakes up the main agent with the results.
00:01:06We can also check our running agents using the tasks command.
00:01:10Under the hood, the background agent works as a separate process that is able to execute
00:01:14complex multi-step tasks autonomously, meaning it can independently handle full-scale tasks
00:01:20on its own.
00:01:21They allow you to continue interacting with Claude Code's main thread without blocking
00:01:25their execution, so you can use your time effectively instead of waiting.
00:01:28When they are done executing, they report the results back to the main agent.
00:01:32Previously, you had to open multiple terminals with separate Claude instances for this purpose.
00:01:37Now Claude does this itself, managing background agents internally, coordinating multiple agents,
00:01:43and integrating the results on its own without the manual setup we were using earlier.
00:01:47They say background agents are ideal for researching while coding, reviewing workflows, performing
00:01:52parallel tasks, and handling any task that shouldn't pollute your main context window.
00:01:57Before we discuss whether these agents actually save us time, let's take a quick break to talk
00:02:01about Automata.
00:02:02After teaching millions of people how to build with AI, we started implementing these workflows
00:02:07ourselves.
00:02:08We discovered we could build better products, faster than ever before.
00:02:11We helped bring your ideas to life, whether it's apps or websites.
00:02:14Maybe you've watched our videos thinking, "I have a great idea, but I don't have a tech
00:02:18team to build it."
00:02:19That's exactly where we come in.
00:02:20Think of us as your technical co-pilot.
00:02:23We apply the same workflows we've taught millions directly to your project, turning concepts
00:02:28into real, working solutions without the headaches of hiring or managing a dev team.
00:02:33Ready to accelerate your idea into reality?
00:02:36Reach out at hello@automata.dev.
00:02:38So I tried out what they were suggesting, and from the time I've spent with them so far,
00:02:42I really like what I'm seeing.
00:02:44I think these agents have a lot of potential.
00:02:46These background agents can handle any research task that takes a lot of time or might bloat
00:02:51your context window.
00:02:52You can simply move that research to the background and assign a separate task to the main agent.
00:02:57This allows the main agent to work on something completely different while remaining totally
00:03:02isolated from the research task.
00:03:04These agents make it much easier to implement changes on separate pages at the same time
00:03:08by parallelizing tasks.
00:03:10Instead of waiting for one agent to handle everything sequentially, you can update them
00:03:14simultaneously.
00:03:15For example, if you're switching all your components to a new library, you can use background
00:03:19agents to update them all at once.
00:03:21Other than that, you're often left with tasks like reviews that take up a lot of time.
00:03:25You can use async agents to launch several of them simultaneously, one for a security
00:03:30review, one to identify performance bottlenecks, one for refactoring code, and another to check
00:03:35for duplications or unused code.
00:03:37You can run all of these tasks in the background while your main agent stays focused on a completely
00:03:42different feature update.
00:03:43This way, each agent works separately and handles its own task independently from the
00:03:48others.
00:03:49Now here's the unexpected angle I mentioned earlier.
00:03:51You can even equip background agents with the ability to manage their work in a Git work
00:03:56tree, allowing them to commit and push changes in isolation.
00:03:59Some of you might have already thought about this, but for those who don't know, Git work
00:04:03trees are an extremely underutilized feature that lets you work on multiple branches in
00:04:07separate directories.
00:04:09This means you can avoid constant branch switching in a single workspace and simply spawn as
00:04:13many agents for as many features as you like.
00:04:15To use this workflow, I set up a custom slash command that creates a background agent to
00:04:20work on tasks in separate work trees.
00:04:23All the instructions for this command live right in the commands folder, which you can
00:04:27find inside the .claud folder.
00:04:29Let's say I'm working on an app, but I'm not happy with the theme.
00:04:32I have 3 different theme ideas, and I want to implement all of them to see which one
00:04:36looks best.
00:04:37Without work trees, I would have had to clone the repo separately, create 3 different branches,
00:04:42or open separate .claud instances just to keep the tasks apart, but with background agents,
00:04:46all of that changes.
00:04:47The 3 agents worked in parallel to implement those themes separately, all while leaving
00:04:52my main design completely unaffected.
00:04:54This allows me to simply merge the theme I like into the main project and discard the
00:04:58others.
00:04:59This has made the implementation process significantly faster.
00:05:02Because all the changes are isolated, I can choose the theme I like best from the different
00:05:07options.
00:05:08It gives me the freedom to experiment and execute without worrying about breaking the main design
00:05:12or having to revert unwanted changes.
00:05:15In the changelogs, they announced some other really cool features, and one that I personally
00:05:19am really thankful for is Instant Autocompact.
00:05:23Compacting used to take a good 2-3 minutes, but now it happens significantly faster because
00:05:27it's likely maintaining histories for each project right in the .claud folder.
00:05:32Claude Code also added a prompt suggestion feature where you can simply press enter to
00:05:36accept a suggestion or just type a new prompt in its place.
00:05:39You can use the agent flag while starting Claude to run it as the specific agent you have in
00:05:44your project, allowing you to delegate tasks to it directly.
00:05:47Claude Code now allows you to fork sessions and choose between them when you launch Claude
00:05:51with the resume flag.
00:05:53All you have to do is press the escape key twice, and your current session will be forked.
00:05:57Now, even though background agents are great, there are certain areas where you shouldn't
00:06:02use them, as they will hinder your workflow instead of making it better.
00:06:05For example, you don't want to background any task that requires your input because that
00:06:10would cause the agent to get blocked and unable to proceed.
00:06:13You also don't want to background any task that requires your approval while it's running.
00:06:17It would just get blocked in the background, and you could actually lose track of it if
00:06:21the main agent is busy with another task.
00:06:23It is also highly recommended not to assign tasks that depend on each other, as this can
00:06:27create conflicts in your agent's work.
00:06:30Instead, it's much better to assign isolated tasks to each agent.
00:06:34There are a few best practices you need to keep in mind to get the most out of this workflow.
00:06:38While using background agents, you need to keep token usage in mind.
00:06:41Even though they speed up your work, each agent consumes a significant number of tokens, so
00:06:46it's important to track this if you want to avoid high bills.
00:06:49Since agents return their output to the main agent, you have to make sure your tasks are
00:06:53isolated.
00:06:54Also, when you're spawning an agent, give it a descriptive name.
00:06:57That way, if you're managing multiple agents at once, you'll know exactly what each one
00:07:01is doing.
00:07:02That brings us to the end of this video.
00:07:04If you'd like to support the channel and help us keep making videos like this, you can do
00:07:08so by using the super thanks button below.
00:07:10As always, thank you for watching and I'll see you in the next one.