Chapter 1: Getting Started with Muse

Most people use AI like a search box. They type a question, read the answer, and start over the next time. That works for quick questions. It does not work for getting things done.

This course is about the other way to use AI: hiring an agent. An agent is an AI that can act, not just answer. It can check your email, fill out a form, shop on a website, or build a small app, and it can keep working on a goal over days and weeks. Like any new hire, it does its best work when you tell it clearly what you want, check what it does, and teach it how you like things done.

The skills in this course apply to any AI agent. We practice them with Muse, Meta's personal AI agent. You do not need any experience, and you do not need to know how to code.

1.1 What this course is about

Here is the difference you should notice by the end.

Before: you ask one-off questions, you check your email and repeat boring tasks by hand, you either trust what the AI produces without checking, or never hand it anything, and you think building an app means learning to code first.

After: you give an agent clear instructions that say what you want and how you will know it is done, you work toward a real goal over several weeks, you have a daily check running, you hand off shopping and website work and approve the important steps, you review a draft before anything is sent, and you build and share a small app with a working link.

You will leave with five things to show for it:

• Muse set up and working for you
• a completed course goal
• one daily check that runs on its own (we call it a daily watch)
• one approved draft routine
• one app you shared with other people

Each chapter follows the same rhythm: you read, you try the move in a simulator where nothing is live, and then you do it for real in your own Muse.

1.2 Which agent to use

Muse is not the only AI agent. GrokBot is another one, and you are welcome to use it if you already do or prefer it. The skills in this course (giving clear instructions, building trust, approving, drafting and reviewing) apply to any agent, so nothing you learn here is locked to one product.

We use Muse for every example in this book because it is free, which makes it the easiest for everyone in the course to get started with. Screens, steps and simulators all come from Muse. If you use GrokBot, expect to translate the examples to its screens, and look for the same ideas under nearby names.

1.3 Setting up Muse

Install Muse on your phone and sign in. Muse works like texting a friend: you talk to your agent in a chat, and the work it does shows up right in the conversation, such as documents it filled out, reminders it set, and follow-up questions it asks.

Take a moment to look at the five tabs along the bottom of the screen. You will use all of them in this course.

Tab	What it is for
Chat	Home base. Talk to your agent and see its work right in the conversation.
Feed	Posts your agent schedules for you.
Ideas	Things your agent offers to take on, waiting for your yes or no.
Goals	What you are working toward, what your agent is tracking, and progress.
Library	Documents and other things your agent made, collected in one place.

Tap your agent's picture to open the status screen, which shows what your agent has done (Activity), what is waiting for you (Approvals), and what is scheduled (Upcoming). Settings is where you manage Connectors (the apps you let Muse reach, such as email), Wallet (how Muse pays for things), the secure credentials store (a locked place for your passwords), Permissions (what Muse is allowed to do), Devices, and Data controls (your privacy choices).

1.4 What to know about privacy

Before you hand an agent real work, it helps to know how your information is handled. Here are the facts, so you can decide with a clear picture.

• Your chats do not have to be used to train AI. You control this with a switch in Settings, and the next section shows you where it is.
• Muse cannot see the sign-in details you give it. Passwords and similar details go into Muse's secure credentials store, and the agent itself cannot read them.
• Your usage is not used for ads.
• Your work in Muse is not necessarily confidential. Muse does its work on its own computer, which lives online. It uses that computer to browse websites and do tasks. The Muse team can still get into that computer, and they may look at it to find and fix problems. Meta's policy calls this computer a virtual machine.

The last point is the one to keep in mind. Treat Muse like a capable assistant at a company that has a support team: it is safe and useful for everyday work, and it is a good idea not to hand it anything you would never want a support engineer to see. For almost everything in this course, such as email summaries, shopping, and building a small app, that is not a concern.

Meta's full details are in the Muse privacy policy. It is worth reading once, and policies can change, so check it again from time to time.

1.5 Turn off sharing your chats

Muse can use your chats to improve AI at Meta. That is a choice, and you can say no before you tell it anything personal.

Open Settings, then tap Data controls. Find the switch labeled Help improve our AI models. The line under it says what it does: it allows your interactions with Muse to be used to develop and improve AI at Meta. Turn the switch off. Off means your chats are not used that way. If it is already off, leave it off.

The same screen lists archived side chats, a way to import memory, and a way to download your agent data. Reset is different. It is shown in red, and it permanently deletes your Muse data, including chat history, files, and active tasks. Leave it alone unless you mean to erase the account. The name on the button is your agent's name.

The phone below is a practice copy of Muse. Press Play and it opens Settings and turns the switch off. Nothing on it is your account, and you can pause and tap around yourself. A later section explains these lessons in full.

Simulator: Data controls. Open Settings and turn off the switch that shares your chats.

Now do it in your own Muse. Open Settings, open Data controls, and turn Help improve our AI models off.

1.6 Your main chat and side chats

Most of your work with Muse happens in one place: your main chat. Think of it as texting a secretary. You send a message, and your agent gets on with it and reports back. Because it is one long conversation, your agent builds up a sense of you over time.

You can also create side chats. A side chat is a second, separate conversation, so a topic does not get mixed in with everything else.

In this course, the book's running examples live in side chats. This chapter's setup happens in your main chat. Starting with Chapter 2, create a new side chat for each chapter you work on. Name it after the chapter, for example "Chapter 2: Boring Stuff". Keep that chapter's exercises in it. That keeps the course work together and out of your everyday chat, and you can look back at a chapter's work whenever you like.

Your main chat stays yours. Use it for your real, everyday requests, which brings us to the most important advice in this chapter.

1.7 Use Muse every day, not only for class

The assignments in this book are the starting point. The real learning happens when you use Muse for the rest of your life too.

Each time you use Muse on something you actually need, such as a reminder, a message to draft, or a question about your day, you practice giving instructions, checking and approving without a lesson attached. That practice is what makes the skills stick. People who use their agent a little every day are far more comfortable with it by the end of the course than people who only open it for assignments.

You do not need a big plan. Each day, hand your agent one small, real thing, and notice how it went. If you are not sure what to try, open the Ideas tab and see what Muse offers.

1.8 How the simulator lessons work

Each lesson has a simulator: a practice copy of the Muse phone. You can tap anything in it. Nothing is your account. The people, inboxes, and cards are made up.

Press Play under the phone. That lesson hands Muse a script, then does the moves and explains each one. A different lesson can hand Muse a different script, so the same phone answers differently. You can pause and look around: open a tab, send a message, open the sidebar. Restart puts that scenario back to the start. Then do the same thing for real in your own Muse.

Simulator: First Five Minutes. Send a first message and visit each tab.

When you finish, check that you can do these in the simulator, and then in your own Muse:

• send your agent a message
• give your agent a name
• open each of the five tabs and say what it is for

1.9 Set your goal for the course

A goal is the first real thing you will give your agent instructions about, so it is worth doing well.

A weak request is a question: "Can you help me with my email?" A good set of instructions gives the agent an outcome (the result you want) and says what done means (how you will know it worked). We call this a brief. For example: "By the end of the course, I want my inbox to be handled without me checking it every hour. Done means I get one summary each morning, and any reply that needs to go out is waiting for my approval."

Pick a real goal you want to reach during the course. It should be something you care about, small enough to make progress on in a few weeks, and clear enough that you will know when you are finished. Good examples:

• Spend less time on email and still miss nothing important.
• Get my small business's follow-ups out on time.
• Build an app that helps my club, class or customers.

Write your goal in one sentence for the outcome and one sentence for what done looks like. If you cannot say what done looks like, the goal is not ready yet. Make it smaller.

Simulator: Set Your Goal. Create the course goal in the Goals tab and watch your agent start tracking it.

Now do it in your own Muse. Open the Goals tab, create your course goal, and look at what your agent starts tracking. A goal is not just a note. Your agent keeps an eye on it and shows progress, which is how you will check in on it across the course.

1.10 Ideas and the Feed

Muse does not only wait for you to ask. The Ideas tab holds things your agent offers to take on, written for you. You can accept an idea, which starts a short job, or dismiss one that does not fit. Dismissing is a perfectly good choice. You are the one deciding what your agent works on.

The Feed holds posts your agent has scheduled for you. Anything your agent makes ends up in the Library, so you can find it later.

Simulator: Ideas and the Feed. Accept one idea, dismiss one, read the day's Feed posts, then find the result in the Library.

1.11 Write your "About me and my goal" paragraph

An agent that knows nothing about you has to ask about everything. An agent that knows a little can start well.

Muse can remember how you like to work. A standing preference is something you tell your agent once and it keeps using. You will add more of these in later chapters.

Write a short paragraph, five or six sentences, that covers:

• who you are, and the work or business you do
• the goal you chose for this course
• how you like to be asked before your agent acts, for example "ask me before anything is sent or bought"
• anything your agent should never do

Then tell Muse to remember it as a standing preference. Keep the paragraph somewhere you can copy it from, because you may want to use it with another agent someday.

1.12 Your Learner Profile

This book keeps a learner profile for you. It is a short message that you paste into an AI chat. It says what you are working on and how you want your agent to work with you, and it grows as you learn each skill in the course.

Use the learner profile panel on this page to copy the profile as it stands now. It includes where you are working (you are a beginner using Muse) and what you know so far. In this chapter, you learned your way around Muse, how to turn off sharing your chats to improve AI, how to give instructions that name an outcome and a definition of done, how to set a goal, and how to save a standing preference.

The profile and your "About me" paragraph do different jobs. The profile is the same for every reader and covers agent skills and rules. Your paragraph is personal and lives in Muse, where it is saved as a standing preference.

Paste the copied profile into your main chat and ask: "Using what you know about me, what is a good first step toward my goal?" Read the answer and see whether it fits.

1.13 Summary

You hired an agent: you set up Muse, learned the five tabs, turned off sharing your chats to improve AI, and learned how simulator lessons work. You wrote your course goal as an outcome with a definition of done, saved a paragraph about yourself as a standing preference, and copied your first learner profile. In the next chapter, you put Muse to work on the boring stuff.