The Real Agent Experience: What Working with AI Actually Looks Like

Andy Sack And I kind of freak out, but I'm not that freaked out. It's not quite like losing the thesis in college because it's really just been a day's worth of effort and work that I wouldn't have gotten anything done because I had meetings out of the office anyway. So I'm kind of annoyed and I start arguing with the agent saying, what are you talking about? We went step one through four.

I know you have those files. And it's like, no, they don't exist. And I basically had to drive home, go to the laptop that was at my office, and basically spend 90 minutes helping and commanding, telling it that it had to find those files step by step. And sure enough, I was able to find those files and...

as I told Adam, unfuck itself. This is AI First with Adam and Andy, the show that takes you to the front lines of AI innovation and business. I'm Andy Sack and alongside my co-host, Adam Brotman, each episode we bring you candid conversations with business leaders transforming their businesses with AI. No fluff, just real talk, actionable use cases and insights for you.

I've got an actionable use case. Today, we're going to talk about my own usage of Claude Cowork and agents. And I'll share my use case. And hopefully, what you'll get from this is a reason to take a project of your own and start using Cowork and how this might apply to you in your business.

So with that, let me share a story from yesterday. I have been working for the last two plus weeks on a project for Forum3 with Adam, and we're going to share the details of that project with you shortly. It's a writing project that involves a long document. Yesterday, I woke up at eight o'clock in the morning and I started working with co-work.

I had a bunch of meetings out of the office and so co-work now has a feature that is called dispatch and it allows you to use your mobile phone and basically talk to the co-work that's running on your laptop, your MacBook computer. So I go to these meetings starting at eight o'clock in the morning. I have a coffee meeting. I have a lunch meeting.

And then I have another coffee meeting all out of the office. As I'm going from meeting to meeting to meeting, I'm pulling up to the traffic light and literally pulling up my mobile phone and talking to dispatch verbally saying, hey, can you tell me the status of that task? How's it going? And then I drive up to the next traffic light and I'd look at the response like, looking at a text message and it would tell me and I would be like, yeah, proceed with that.

Go ahead and make those edits. so literally as I'm going meeting to meeting, I'm advancing and editing this document at a rate that I couldn't have done it if I had been sitting in front of my computer alone. And I am invigorated by doing this and I feel like I'm living in the future. we had outlined in the morning a six step process.

Around two o'clock, I'd gotten through step four and I said, okay, I'm ready to take step five. And the cowork messages me on dispatch on my mobile phone. I'm looking at it and says, oh, the files that you've been working on basically all day don't exist. And I kind of freak out, but I'm not that freaked out.

It's not quite like losing the thesis in college because it's really just been a day's worth of effort and work that I wouldn't have gotten anything done because I had meetings out of the office anyway. So I'm kind of annoyed and I start arguing with the agent saying, what are you talking about? We went step one through four. I know you have those files.

And it's like, no, they don't exist. And I basically had to drive home, go to the laptop that was at my office, and basically spend 90 minutes helping and commanding, telling it that it had to find those files step by step. And sure enough, I was able to find those files and... as I told Adam, unfuck itself.

And I was able to recover the files after 90 minutes. And the process was insane. Because as I mentioned, I felt like I was working in this totally new way in the future. I was invigorated.

And then I had this crappy mess of a situation and I had to get unfucked. And I told Adam that story this morning. was like, oh, that's what we should lead the mini episode with. So Adam, there you go.

What do you have to say? Adam Brotman I love that story for a couple of reasons. One is, because I it's really instructive for people who have not been working with agents on any level or tried working with agents. So A, what is an agent?

the easiest way to think of an agent is we're talking about an AI system that's not just a chat bot. Most people think of AIs as chat bots that are kind of slightly better Googles. maybe can help you write something. They really think of it as just that.

Like, this thing can write things, whether it be an answer to a question I have, so I like it better than Google, because I don't have to fish through links, or it helps me write an email, or whatever. Maybe it can summarize a document. So that's how most people think of these chatbots. But imagine if that AI also had access to computer tools, like Search and document creation and presentation creation and it could do like multiple steps of things and could actually do work for you like actually got things done for you and so it was it wasn't a chat it was a task and in fact so when you think about agents like open claw or Claude's Cowork, which is Andy's example or even codex agent desktop agent from ChatGPT and the like Like that's what we mean by agents.

They can get things done for you And it's like Andy said, they can be exhilarating, they're powerful. But like Andy just showed, they can screw up and they're more powerful. They get more done for you over a longer range time, but they can make mistakes. And number one thing we want you to hear from the story is remember that when we tell people like, man, you got to try co-work.

We're also saying you got to be careful, like what documents, what systems you give it access to. You got to learn about these things and be responsible, number one. And Andy's story is instructive of that. But once you sort of get your head around that and you're like, okay, I'll be careful.

I'm willing to try it and I'll be careful. And I get the dangers of like, you know, the things that can go wrong. if you start using them, you also get the same experience Andy had, which is like, wow, Andy's giving, delegating this writing and research and editing all tasks to this agent on his computer back home while he's driving around having three coffee meetings and lunch. and He's having dialogue with this thing and I told Andy and tell me Andy you can chime in I said Even with the 90 minutes you had to go to go fix things and even with like the frustration of it all like in that day You were able to get done on our project X amount of work which is probably ten times more than you would have gotten done if you had just taken the time you spent Working with the agent and instead of working with the agent try to just do the writing and the editing and the research yourself.

And that's pretty reflective, I think, of a lot of agent experiences and why people who actually use them and see the mistakes they can make. Because part of the issue with the mistakes also is they teach you what the systems are capable of. They teach you to be more careful in the future. They also teach you how to be a better user of the agents, right?

Andy Sack 100%. Adam Brotman And so you get better and you up skill yourself. And I want you to talk about your skills in a second. Like you up skill yourself and you up skill your agent when you work with them and you're 10 Xing yourself at times.

And it's kind of amazing, but it's unlike working with a human and it's unlike working with a software system, which are the only two patterns or analogs we have right now working with either technology or people. This is something new. It's something weird. It's something amazing and it's something dangerous all at the same time.

And It's hard to explain until you use it. And I'm glad you told that story, because I think it's a good example for people who haven't been working with agents of how kind of amazing they can be and how different they can be from what they're used to. Andy Sack Yeah, I mean, the thing that you sort of got a glimpse from when I tell the story is just how invigorating it like I talk about how it felt like playing a real life video game. And, the notion that I'm waking up and first thing I'm doing is logging into project.

There's a excitement, enthusiasm, thrill from the video game of working with agents in real life. that I just hope people in the audience get to experiment on whatever project they're working on. Adam and I. In our book, we talk a lot about AI first leadership and the AI first mindset.

That intellectualizes. The story I just told you and the visceral invigoration, the emotion that comes with it and how fast and productive it was while I'm driving on it. Like I felt like I was living in the future. And I hope that the people listening to this take a project and work with that.

The point that Adam was asking me to share with all of you is that in the process of doing this project and working with Cowork over the last two to three weeks, just to orient you to my own agent usage, I started using Cowork and had the aha moment in January 9th. I remember the date. I was out of the country. And then I have set up and I have two agents in OpenClaw.

So I'm not a novice. Like I'm a... Two to three weeks ago, I was an advanced beginner as an agent user. In the last two to three weeks, I went from an advanced beginner to an advanced intermediate agent user.

And how do I know that? Because I went on using Cowork two weeks ago. I had five skills. And today, I have 28 or 29 skills.

And a skill is the Adam Brotman Explain what a skill is. Andy Sack technical instruction that I create with cowork that that I can call at any time not from now and going forward. I've basically told cowork this is a skill I want to use and it's created a little prompt that it gets called and reused. Is that did I describe it well?

Adam Brotman I think, yeah, in other words, yes, in other words, as you've been working with cowork, every once in a while, you and cowork will figure out a way for cowork to do something for you that it didn't know how to do until you sort of figured it out with cowork. And at that moment you go, aha, save that, save that thing, that instruction that you just figured out with me as a skill so that the next time I want you to do it, I'm just going to say, invoke skill number 24.

And that's the skill that you already know how to do this thing. I don't have to reinvent the wheel. Andy Sack Yeah. Yeah, so an example of some of those skills are prep for a meeting skill.

Respond to email skill. What do those two skills do? The first one actually looks at my calendar, looks for what I'm having for a meeting with, goes on the web, researches the person in the company, and gives me three bullet points about who I'm meeting with and what the meeting is about. Adam Brotman And you might have, by the way, on that example, you might have done that one time with cowork and it didn't quite do it the way you liked.

And you said, no, I actually like you didn't go to LinkedIn. You just went to the web. I want you to go to LinkedIn as well as the web. And I want you to format it this way for me.

And then you get it right. And then you go, ⁓ save that snapshot. That's the skill. Andy Sack That's right.

When I tell it to write emails, I have to instruct it because I have my personal email and I have a few work emails and I have to tell it always draft emails from this alias. and so like their skills, they're repeatable things that you do every day. And over the last two weeks, I've gone from five to 28 and those are available to me for the next decade. Skills are going to be incredibly important to your individual work.

And they're going to be things that people both share and sell. So organizations are going to be buying special skills and adapting them to their own workflows and processes. In many ways, this is the business that Forum3 is getting into on behalf of restaurant, retail, and other industries. And we'll talk more about it in another episode.

Adam, final comments if you have any. Adam Brotman No, I thought that was a helpful anecdote and I hope people got a kind of a glimpse of what it's like to work with agents, both good and bad. It's thrilling, but you've got to be careful. Andy Sack That was great.

Great. Thank you all for listening to AI First with Adam and Andy. For more resources on how to become AI First, you can visit our website, Forum3.com download case studies, research, briefing, executive summaries, and join our email list.

We also invite you to connect with our AI First community, a curated hub and network for leaders turning AI hype into action. We truly believe you can't over-invest in AI learning. Onward.