Video 1 - What Your Robot Assistant Can Do

This opening video sets the stage for everything that follows. David explains the core idea behind a robot assistant — it's not a chatbot you talk to once and forget. It's a persistent, trained system with three pillars: memory (a workspace of files and instructions it reads every session), skills (documented workflows that tell it exactly how to handle specific tasks), and reach (MCP connections to real tools like your calendar, email, Slack, and the web). The distinction matters: a chatbot is a contractor who forgets everything between jobs. A robot assistant is an employee who learns your systems and gets better over time.

To make the concept real, David runs a live demonstration of his daily briefing — the robot checks his calendar through Fantastical, pulls his task list, scans Slack and email, grabs the weather, and assembles a branded one-page PDF that he prints out and puts on his desk every morning. It's not a pre-canned demo; it's just Sunday morning and the robot doing its job. You'll see the MCP connections firing in real time, the skills being read, and the final output — a professional daily brief that would make any CEO's assistant jealous.

The video closes with a preview of the series ahead and a simple homework assignment: write down a list of the donkey work in your life. The repetitive, tedious, computer-based tasks you're tired of doing. By the end of this course, your robot assistant will be handling them for you. No setup required for this video — just show up for Video 2 ready to learn how your assistant actually thinks.

Video Transcript

Hello, I'm David Sparks. Welcome to the Robot Assistant Field Guide. I can't wait to start teaching you how to build your very own robot assistant. Now, we've been hearing a lot lately from these AI companies about how they're going to cure cancer and change the world. What they're not saying is that they have already changed the world in a way that can matter to a lot of us. And that's in helping with the tedium — the work that we don't want to do every day. Going to websites, filling them out, dealing with email, those kinds of things. What I like to call the donkey work of our lives. We all want to be creative. We want to do the things that we're best at. We don't want to waste our time with the administrative tedium. And that's what a robot assistant can do. It can handle that for you. It can do it right now. I've already built my own robot assistant. It does this stuff for me every day. And by the end of this course, it will for you too. So let's get started.

But before we get into building our robot assistant, let me just talk a little bit about how all this works. A legitimate question heading in is, what exactly can this robot assistant do for you? Well, the answer is a lot.

First, it can access and process information. I'm going to teach you in this course how to set a text bank on your device that it is constantly working with, keeping track of information for you so you don't have to. Humans are great at processing, but we're not good at memory. Your robot assistant will do that for you, freeing you up to do some of your best work.

It can also name and manage files reliably. I've spent 20 years teaching people the best ways and workflows to manage documents and digital files. So many things that were struggles for us throughout that time are now trivial with the assistance of a robot assistant. I'm going to teach you all about that.

It can also work with the web. While at first this seems like a small point, it's actually huge. Any bit of tedium that you have to go and do on the web, it handles for you. Once you give it the right instructions, it does it repeatedly and reliably.

Likewise, the robot assistant can work with applications. Thanks to the Model Context Protocol from Anthropic, an increasing number of applications are tying themselves right into your robot assistant. There are new MCPs every day, giving your robot assistant even more utility.

And then there's the problem we all love to hate — email. With your robot assistant, you can triage your email and so much more. Indeed, anything repeatable on your Mac is subject to automation through your robot assistant. As I was training my robot assistant, I found myself jumping out of bed every morning with a list of ideas of things that I do every day that I don't want to do anymore. With just a little bit of training, my robot assistant was able to take those over for me. I can't wait for you to have that exact same feeling.

The next question is, how does it work? Well, there are three pieces to a robot assistant.

The first one is memory. Using a traditional chatbot, you'd always discover that while the thing is useful, it has a very poor memory. So one conversation to the next, it doesn't really remember what's going on and the context gets lost. You've got to rebuild that context every time. With a robot assistant, your workspace folder holds everything the assistant needs — instructions, style guides, project files, skill documents. This is something we're going to build together throughout the course. And this is the assistant's brain. They're just files on your computer, but the assistant can constantly be going back and reading them, refreshing its memory, and making it way more useful to you.

Now the second piece of this are skills. They're documented workflows that tell the assistant exactly how to handle specific types of work. A content creation skill, an email drafting skill, a daily briefing skill. Each one is a set of instructions you've written or customized from the template I'm giving you, along with the assistance of your robot assistant, to give it the skill to do the task you've asked.

And then the last piece of this is reach. With MCP connectors, the assistant can interact with your real tools — calendar, email, Slack, task manager, notes. All that stuff is available, in addition to the entire web. Without reach, the assistant can only read and write files. With reach, it can check your schedule, search your messages, and take real action.

So we're going to dive into this nuts and bolts in the course starting from the beginning, but before we do, I just want to show you one trick I can do with my robot assistant.

Here I am in Claude Cowork. I've pointed it at my SparkyOS folder. First thing I'm going to do is tell it to remember what's there. This is a little expansion snippet I run every time we start it up, and I'll tell you more about this in the course, but you're having it check its skills and its index so it remembers what generally it can do. Okay, it's ready to start work.

I want to make what I call the daily brief, so I'm going to tell it to remind itself about that. "We're about to create a new daily brief. Please check your skills before we begin." Yes. So a few things — I'm using a text-to-voice tool to speak to it. I find that faster than typing and more conversational.

And as you see, it's now going to work for me. It's going to use the MCP with Fantastical, my calendar app, to get my available calendar events. It's got a list of all my tasks. It's checking those. It's going to look for recent messages in Slack. It's going to check my Drafts Cowork inbox. These are all things I'm going to teach you about throughout this course. It's even going to check my email inbox and the weather for where I live here in Foothill Ranch.

So it's off doing this. And this is a thing when you've got a robot assistant — it will occasionally go off and do other work. The way I handle that is I make it a full-screen app, put it to the side, and I just come back and check on it every once in a while. Right now it's doing, looks like, the last bit of this — checking the weather data. Down the right side of the screen you can see the context, the different applications it's looking at, the different skills and instructions it's reading. This is what I was talking about — memory plus reach.

One of the nice things about the robot assistant is it will narrate what it's doing, so you've got a good idea what's going on. My robot assistant manages a lot of information for me — ongoing projects, tasks, email, calendar — so there's a lot of information for it to collect. And what I'm aiming for here is to have it print me out a one-page daily brief. I've always been jealous of presidents and CEOs who walk into their office and some smart assistant has put everything they need for the day on one page. So I decided to train my robot assistant to do that for me. In the course, I'm going to teach you the same.

Okay, you can see it's getting more active now as it's finishing up the project of making my daily brief for me. And there it is. So it has created my daily brief. Looking through all of the MCPs and all the data that it has for me, I'm going to open that up in preview and there it is — my daily brief. The thing that I put on my desk every day.

Let me put it in the center here so you can see it. Let's just look through it. At the top, I've got the title and the font that I chose. It's got the date and the week number, a little Stoic quote underneath — that's kind of fun. I've also got the weather for today, tomorrow, and the next day. Under that I've got the calendar for today. Today is the gotcha day for my dog, so that's kind of fun. And I've got my events for today listed below that.

I have repeating tasks that I do on certain days. This is the list of the repeating tasks I normally do on Sunday. My robot knows that, so it's reminding me those are my repeating tasks. It's also giving me a list of workbench items. These are the things that are active that I'm working on right now for one reason or another, and I've got a few past-due items here that need updating. I've also got some pinned items here that I need to take care of. It's following those for me because I've pinned them in my system. They show up every day until I get them done.

It's also got my calendar for the next few days — tomorrow and the week ahead — and approaching deadlines with a number list to the right telling me how long they are away. It's giving me an update on my various projects I've got going on. The Productivity Field Guide workshops are about half done and the Robot Assistant is just getting started. I'm recording now — that's kind of exciting.

And then it's also giving me awareness as to notifications, showing me what's coming up in Slack and Cowork and email and SaneBox. So it's just telling me anything that I need to make sure to pay attention to. I'm really bad at checking social media and mail. The reason I added this to my brief is to give me that reminder in the morning in case there's anything that needs my attention.

So every day I've got this on my desk sitting there waiting to help me out, get through the day. It's important to note that what I just did was not a pre-canned demonstration. It was just Sunday morning and my robot assistant built me my daily brief like it does every day.

While I tell you this course is about programming or building your own robot assistant, it's a lot more like training a new employee. You've got to teach it what your preferences are, how you want things done, and when it makes mistakes you have to correct it. Once it gets that locked in, once it's doing these things exactly the way you want, it never forgets it and does it right every time. That's what you're going to get by the end of this course.

Now, why is this possible now? It really starts with Claude Cowork, which is the first frontier model LLM that's capable of having that reach into other applications in your browser and the ability to have memory when you build a memory system like we're going to do in this course. When you put all that together, that gives LLMs so much more power than the usual chatbot that's interesting but doesn't really turn into action.

So we're going to have a great time in this course building your robot assistant out. I have made a starter kit for you that you're going to run as we get through these videos together. So you've got a real big start. There's also a workshop series in this course that we're going to be going through, and they are essential. Because hands-on is the way to really make this work, and I can't wait to walk you through that as well.

Something you can do at the end of this first video, before you go on to the next one — sit down with a piece of paper or open up a text document on your computer and write down a list of the donkey work in your life that you are tired of doing that fits on a computer. If there's something you're doing that's donkey work on your computer, there's a very good chance that by the end of this course, you're going to have conquered it. Your robot assistant's going to do it and you won't have to.