Micah Johnson 0:00

That we realized that AI was so good at inferring processes and procedures that based on all the AI employees that we had, it accurately inferred our functional areas, the order at which they go in from start to finish while working with customers, and within each functional area, the process map for each of those.

SPEAKER_00 0:25

Oh, it is so sexy.

Micah Johnson 0:28

Welcome to the Biggest School podcast. I'm your host, Micah Johnson.

Alane Boyd 0:32

I'm Elaine Boyd. And today we're going to be answering the question of where do all these AI employees, these agents, these skills, where do they all go so that our team members know they exist?

Micah Johnson 0:44

I like how you put that, that they know that they exist because I mean, we ran into that with our own team. We see our clients running into that. You can build the coolest things. And if nobody knows they exist, then, well, what was the point of building them other than for you to use them?

Alane Boyd 1:00

Yeah, Micah, the question that I start seeing when we're working with companies and they're jamming their build-in skills, they're have creating really cool agents. And then they're like, but how do we, now that we have all these, how do we make sure the team knows that they are available?

Micah Johnson 1:18

Yeah, I I hear the same thing. How do we keep track of them? And it's it's like they're out in the wild. Like, how do we keep track of all these things? But when we started looking at this, it was you've got skills that are AI employees, you've got artifacts that are tools, you have N8N workflows. Some of them are behind the scenes, some of them are actively catching web hooks and actually being used. You have scheduled tasks. Like that's a lot to deal with.

Alane Boyd 1:49

And it's too much to remember. I mean, when we first started building skills and we have AI usage Slack channels, so we can share ideas, we can tell each other about it. But it gets to a point, because now we're at 165 AI employees at the company, that we can't retain all that. That's more than we need to put pressure on our brains to memorize. And so we did get to this threshold of like, hey, we can't remember all of the ones that we have that we've built and our team can access.

Micah Johnson 2:20

I mean, I don't know what this says about me, but I couldn't remember some of the ones that I built early on.

Alane Boyd 2:26

I think that says a lot about you, Micah.

Micah Johnson 2:28

I think it does.

Alane Boyd 2:29

You like building, you don't like using.

Micah Johnson 2:31

Yeah, absolutely. I build things. We see so many people like that right now. Like this technology enables you to build stuff. And yeah, there's like we've had episodes on this. There's some like cautionary stuff you need to know when building, but like, don't frown on the building. Like, you still need to experiment, you still need to learn. The days of like, I've got this idea, I need to go to a whole team, that's gone. So that's pretty cool. But we ran into this with landing pages, right? Like, we just did a couple episodes on our system to build landing pages quickly and high-quality landing pages, we think. Um, they're working. So there's that. But we built so many that even I was like, what was the URL for that? And you were like, Do we have a landing page for this? So we built a landing page tool that auto-indexes everything that we produce and publish and gives us an interface for that as an internal tool. And so today we're gonna talk about the AI employee org chart, which does the exact same thing, but for all of our automated processes.

Alane Boyd 3:39

Oh my gosh. And it's so cool. And it's pulling in from not just our cloud skills, but it's also pulling in from our N8N, where we have all the other AI agents built out too.

Micah Johnson 3:50

Yeah. I would say there's a few things that really surprised me as we built this out, like a few aha moments, because it's still hard to get out of the thought process that we were in from, I would even say, old school software building. Right. So that first glance, it's like we just build this search function. And search is important, but one of the things that happened as we built this tool out was that we realized that AI was so good at inferring processes and procedures that based on all the AI employees that we had, it accurately inferred our functional areas, the order at which they go in from start to finish while working with customers, and within each functional area, the process map for each of those.

SPEAKER_00 4:45

Oh, it is it is so sexy.

Alane Boyd 4:47

And when we built this, I was just so tickled. It's so cool looking. And how I put it by department, how you can filter by department, you can search. I the and we get to a point, and I know other companies are also running into this, is that you you might have AI employees in multiple platforms, just like we do, where we have them in in Aiden, we have them in Claude, you might have some built out in Grokbot, you might have some in Chat GPT. And so it gets to be like, well, where where do we have what? And what, and so we're not having to rebuild and redo. We can access things as long as we still have those accounts, we can still use them because we still use both in Aiden and Claude skills.

Micah Johnson 5:31

Yeah, there's a purpose for both of them for sure, and a use case for both of them. And even, you know, now we're getting into, I don't know if we're going back to the AI assistant trend or getting into a new AI assistant trend, but the GrockBots, the Dots, the Muse, the Claude duets, right? All of these types of systems now introduce a whole new layer to this. I guess what I would say, Elaine, as we kind of talk through this, is documentation today has evolved into tooling. So instead of just creating a spreadsheet of all the AI employees, that was what we used to have to do. Now it's fully interactive. It's a tool, it's an interface, it's powered by AI. So one of the things that we added was if you're looking for an AI employee and it doesn't exist, well you don't have to stop there. We added to the tool that it would allow you to actually say, okay, need help building a new one. And it would search the database, make sure that nothing ever like doesn't already exist, and then come in and help you draft up a whole brand new AI employee.

Alane Boyd 6:44

Oh my gosh. It was already so good. And then whenever I did that piece, I was like, oh my gosh, I didn't know I could love this more. Like this is so cool. And it'll go ahead and draft it. You can copy the prompt and put it in wherever you're building. So it took off so much work. And Mike, I want to say doing it this way was so much smarter and easier, where we just said, hey, these are this is access to the skills and in AidN that we have built us this art chart, org chart that I'm ha where I ran into so much trouble and eventually would just give up every time is when I was trying to do human first. These this is our human org chart, and I want you to create the AI employees under that. And I just it became really stressful for me, and I would just eventually kind of give up and abandon it. But now it's like, oh well, now we can easily see where these AI employees fall under the human.

Micah Johnson 7:38

You what's interesting about you saying this, Elaine, is it made me remember one more piece that we put in here. You know, some people are spreadsheet people, and if they see a list in a spreadsheet, it works really well. Some people are Kanban people, some people are process map people, some people are calendar, like right, you have all these different ways that individuals resonate with how things are displayed. So what you're describing is an AI employee org chart. How does that fit into our human-based org chart? How are we looking at all of this information? But you want to see it in a specific way that helps you personally. Whereas I might want to see it in a more like a list format or something else that you know you might look at and go, oh, I hate this. This isn't resonating with me. And there's nothing wrong with that. What's amazing is this build stage is so reduced in time, it, you know, it does not take a lot of effort anymore. That if we want it in an org chart view, it's an org chart view. If we want to have a toggle that goes list view, it's a list view. If we have a card view, whatever we want to do, we can get it. And that's why building things as tools versus just documenting, I think is the big power step forward right now.

Alane Boyd 8:54

Oh, yeah, definitely. And, you know, we talked about it for a second and I wanted to touch back on it is that it is interactive. So when we're looking at our org chart of all the AI employees, that you can click on it and it opens up and gives a summary of what that AI employee does, the connectors that it has, all kinds of information right there. So you can say, oh, is this the one I want, or is this the one that I want? And so you can do that without having to leave the little um dashboard that we created.

Micah Johnson 9:28

Yeah, absolutely. Now we talked about this in our previous episode, but we obviously built this in an artifact. Um, it has buttons to launch the actual AI employees directly in Claude. Um, but you can integrate this with other artifacts. So this is like the perfect setup to say, hey, if somebody needs a new AI employee, we don't want them just building it right away. We want them to make a request to our AI committee. Perfect. Set that up in your AI org chart. And all of a sudden you have the capability to go, do I have an AI org chart or do I have an AI employee that will solve this problem for me, or do I need to request one being made? You know what I'm realizing, Elaine, as I'm saying all this out loud, this really overlaps what we did with our last business with the knowledge base. Ask the knowledge base first. If you have the answer, you can move forward with it. If you don't have the answer yet, you find the person who's in charge of that body of knowledge or that functional area in the business, get the answer, add the answer to the knowledge base. So the next time somebody asks a question, it's in there.

Alane Boyd 10:38

Yeah. Oh God, I love that. And some of this, you know, they can start on their own, even if it does need to go to an AI committee, because we are seeing that structure in place where, hey, there's team members that are trying things, building things, betting that it works, and then it goes to an AI committee to almost um, I don't want to say sterilized because that's a horrible word, but oh, where that it'll work for other team members. It's not gonna just work for Micah or it's not gonna just work for Elaine. If it's a call debrief skill, it's not gonna go in and assign all the tasks to Elaine. You know, it's where they can make it repeatable across the company. And which is great because that individual can start using it and start creating that AI employee. Find that it works. I mean, we have team members on our team that do that, and then they bring it to us and they're like, hey, this is phenomenal. It's saving me so much time. This is the use case. I think other people in the organization could use it. And we're like, oh my gosh, yes, let's make it applicable to everybody else and make it available as an organization skill.

Micah Johnson 11:41

Yeah, I love that. I love that. I guess you could call it sterilized, standardized is probably much better word for that. Um, I do want to like just bring up the sake of training around this because you know, I really love thinking about things in the sense of a scalable process. And so with this, if we have an AI employee org chart, it becomes a tool. The training around this becomes really easy. Everybody just needs to know the one place where to go to see if an AI employee, an NAN workflow, anything that may have already been built so they don't have to reinvent the wheel, it's just one place. Super easy to train on. That makes it scalable.

Alane Boyd 12:25

I want to hear your thoughts. At what point is a skill ready to move on from just an individual to a company? So I'm noticing this kind of tendency to need things perfect. And, you know, we don't feel like, oh, we can make it an organizational skill because it's not perfect yet. And so at some point I see this starting to drag on. And I do, you know, we do want things to be standardized and usable and not waste time trying to fix things. But at some point I start to see what needs to be perfect and what is, if you can't get there, I think that it's overprocessed or you're trying to do too many things in the process.

Micah Johnson 13:07

I totally agree with that. Tide, it's hard to say because uh one simple task might be easy to get to perfect, and then another task is like, ew, we're gonna spend the last eight hours trying to go from it's really good to perfect. And so you want to obviously avoid just spinning your wheels at that. And I, you know, I think the concept of making a request or submitting an AI employee that you built for uh standardization and uh getting other people to look at it, that's gonna help you get to perfect a lot easier, or at least very, very good enough. Because there's still gonna be manual parts, there's still gonna be manual things that you need to do. That's still gonna be incorporated into the system. And I guess definition of perfect is based on who's listening to this. But if it's perfect is 100% automated and I never have to think about it again, then you're never gonna get there. That's not how this works.

Alane Boyd 14:05

Yeah, yeah. Definitely looking for the iterations. And, you know, where's the 80% time saving so the 20% can be the human in there? So definitely balancing, you know, you don't want wrong information happening in these skills. You want it to be correct. But at what point, you know, is it usable and does it help save time for the rest of the team?

Micah Johnson 14:25

Yeah. And I mean, you we can run the numbers, and we've done this with clients, we've done this ourselves. Is like, geez, even if you're taking a 100% manual process and you automate 10%, you automate 20% and you get an AI employee to do that for you, that's 10 to 20% that you had to do manually before. There's not a business case in the world that wouldn't go, oh yeah, let's just throw away that 20% gain because it's not a hundred percent. Like, no, we're definitely doing at least the 10, at least the 20%. Like, let's get some small wins here.

Alane Boyd 15:02

Yeah. And Micah, I think let's say succinctly, what we did here is that we created an artifact in Claude, I guess just in Claude Cowork now because everything's moved or into the Claude Desktop app, because everything's just kind of merging into one Claude. And we created an artifact in there with our design system. So it's all looks beautiful, biggest goal. And we created a searchable org chart of our AI employees that's pulling from Claude Skills and N8N and is interactive. So, like that is everything packaged up that we did.

Micah Johnson 15:39

Yep. So, Elaine, let's talk about what happens or what are the potential obstacles that you run into for our, we need a sound a soundboard for our like dun dun dun dun dun BO report.

Alane Boyd 15:52

For the BO report.

Micah Johnson 15:53

Yep. The biggest obstacle you can run into while implementing something like this.

Alane Boyd 15:57

Well, silos, honestly, if your team is not making organization skills or you're not pulling in, you know, everywhere where you're creating these AI agents and you're just letting each person work independently, well, then you don't have the access that you need to create an artifact or create some type of interactive dashboard that you would need is based on what that person has access to that's building. And so you want to make sure that these AI employees are getting shared because that's what's if it's saving one employee time, it could be helping somebody else. And then you want those other team members to know because at the end of the day, we're trying to build scalable systems, hopefully, at our company. And so if you have access to it, also, Micah, thinking about turnover. Employees do not stay with you forever. So that AI employee needs to be shared with the organization so that if that employee human employee leaves, you still have access to that AI employee. Love it. Micah, what do you think about dropping in some screenshots of the artifact that we built for our AI employees in our community at your.bigasgold.ai.

Micah Johnson 17:04

Yeah, we could definitely do that. We'll do a whole um post on all of this.

Alane Boyd 17:09

Yeah. So if you're listening and you want to see what the heck we're talking about, then come join us. We'll do a post and and share exactly what ours looks like. And that way you can be inspired to build your own.

Micah Johnson 17:21

Awesome. Again, that URL is your Y O-U-R.bigascoal.ai. Free to sign up and check it out.

Alane Boyd 17:28

And quick shout out to our listeners that have left us a review on Spotify and Apple. We appreciate you so much. And if you are an avid liv, an avid listener of our podcast, we would love if you went and did the same. We are so excited to continue doing this podcast for you.

Micah Johnson 17:46

All right, everyone. Thanks for listening to our podcast today. That's it for how to build AI employee org charts.