Micah Johnson (00:01)
People tell us they're scared to use AI because it makes mistakes. But does that mean those same people are scared to use humans because they make mistakes?
Alane Boyd (00:13)
So Micah, instead of kicking off the episode and talking about AI today, let's start with talking about human error first, because we make so many mistakes and it's honestly exhausting just trying to keep up with how to keep those mistakes getting to clients and other people.
Micah Johnson (00:33)
Yeah, it's interesting because as we are prepping for this episode Alane, you and I were thinking back on the different things that we've built within our companies, the different systems, the different processes, and 100% of them have to account for mistakes. And so it's weird to me to hear people saying like, I can't use this tool. I can't use this technology. I can't use this. I don't trust it because it makes mistakes. And then they throw a boatload of money at
salaries for people who make mistakes.
Alane Boyd (01:05)
I mean, I think about conversations even with my son and I'm constantly saying it's okay that we make mistakes, humans make mistakes. It's about what we do next that's important. And it's not conversations about how AI makes mistakes. It's about how humans make mistakes and we need to set in some parameters to make sure, hey, we have the space for that. And we also have some oversight to make sure in business that those don't make it too
the public view, which is the client or whatever.
Micah Johnson (01:36)
I feel like this is already a therapy session for Claude. It's okay Claude, It's a safe space here. You can make mistakes, but I pulled some stats, Alane, that I thought would be kind of interesting to go through on human error, just to illustrate how much we're accounting for human error in our normal processes and normal business.
Alane Boyd (01:47)
Okay.
As long as it's not trivia for me, I do not like those games.
Micah Johnson (01:58)
Oh, okay. Let me rewrite this section real quick. No, just kidding. All right. So 80 to 90% of workplace accidents are just kidding. It's caused by human error. I was going to give you the trivia question.
Alane Boyd (02:02)
Okay.
Micah Johnson (02:13)
So like, of course, right? What else would it be? It's humans making mistakes causing workplace accidents. That's why it's called an accident. ⁓ Let's look at something less dangerous data entry professionals, 3 to 4% average error rate per field. Yeah, per field. That's per field they're entering in not per field they're working in like.
Alane Boyd (02:15)
Mm-hmm.
Mm-hmm.
Okay.
What?
Micah Johnson (02:39)
healthcare or whatever. So if we do the math, out of a 1,000 invoices sent, that's like more than 200 errors that could happen. And with our previous company, we sent so many invoices, we did so much stuff manually, we had to double check there were still errors. I know there's so many companies that are sending invoices with so many errors. It's the reality.
Alane Boyd (02:41)
Right.
yeah.
Oh my gosh, like I think of two examples from our previous company that we built up. One of after we did the merger and we we had 600 clients, we were writing over 2,000 social media posts per week. And I remember one of the executives counted how many mistakes each social media content writer had. And I'm going, wow, that's a really
dick move here and one, but number two is, know, sometimes these calculations on the human error side, like obviously we don't want a ton of errors, but also we can't nitpick on every single error or we wouldn't have any humans working for us. And the thought that I had is like, wow. And I remember the one that had the highest amount of errors in her writing.
also wrote four times the amount of social media posts than any other content writer. So she was really, really fast and maybe had closer attention to detail. Well, that didn't mean we didn't want to do 4x the speed. We just put in another layer for a quality control person to go over her post. So we built in that layer for the human error to be checked. And then the other one is actually about my mistakes. I make a lot of mistakes because I don't need to be perfect.
Micah Johnson (04:21)
⁓ you make them?
Alane Boyd (04:26)
but one of them is actually around this invoice thing. I remember with one of our corporate contracts, we had 200 clients under that corporate contract and at contract renewal time, they needed a spreadsheet of everything, every one of those sub clients purchased from us. Every time I looked at that spreadsheet and I gave my brain a break, so it could be later that day, it could be two days from now, I would find a mistake and have to correct it. That didn't just happen one time, Micah.
I can probably name 10 of those times, even though it's been over a decade since I did.
Micah Johnson (04:59)
Yeah, I mean, what you're describing, that's the reality. That's what we have to deal with as business owners, as founders, as managers, even as system designers, we have to account for human error. We have to account for these issues. And, you know, sometimes they are incompetent people that are making these issues and these errors.
Alane Boyd (05:11)
Mm-hmm.
right.
was incompetent. I should have been let go.
Micah Johnson (05:25)
man, now you tell me.
But there's so many times like what you're describing where there's the person that was created the most errors when writing social content. She wrote 4x the amount of anybody else. She was not incompetent. Her writing was fantastic. In fact, she went in to write books. She wrote novels. She was a fantastic writer. We accounted for those errors through a system.
Alane Boyd (05:43)
She was so creative. Yes.
Micah Johnson (05:53)
And so if we were to take the same kind of perspective that we hear a lot of people talking about with AI right now, and this isn't us saying you have to use AI. This is let's try to change our perspective because if you're too close-minded on certain things, then you're not able to get into the actual benefits of the tools and the technology that other people will be able to and therefore you won't be competitive enough.
Alane Boyd (06:01)
Mm-hmm.
Micah Johnson (06:21)
If we were to take this same perspective, Alane, say, nope, you make mistakes. We can't use you. You said it earlier. We wouldn't have a single employee. We'd have to fire ourselves.
Alane Boyd (06:33)
Mm-hmm. Even just starting off with our company, Biggest Goal, and then this podcast, Automate Your Agency, our company is almost seven years old now. So we were doing this before AI was a thing. And really just talking about automation because of the benefits of having this in place rather than depending on humans to do every single piece. We're inconsistent.
We cannot remember everything every single moment. We make errors. So where can the tool help improve the system but still have the human oversight? That hasn't changed with AI. AI actually makes us better humans because we can be the humans looking at the judgment, the strategy, those things where we can still be that oversight piece. We can catch those errors.
Micah Johnson (07:20)
Yeah, I mean, I think about cross-referencing. That sounds weird. I don't think about cross-referencing a lot. I think about it in the context of system design and being able to execute. And you're bringing up such a critical point on consistency, which is when systems are designed correctly, whether it's human or AI, consistency is the output that we're looking for. We want the outcome to be consistent so it's predictable.
And if we say, okay, how do we make human output consistent and predictable? We have to build systems around that. That's what processes are. That's what SOPs are. That's essentially what's happening in the AI world. Skills are SOPs for AI. Why? Because we need the consistency. That's a building block of a system to produce that. And that's what we need to look at.
Where you're going with this is, we've got this whole concept of it's either or, it's mutually exclusive. I'm doing AI automation or I'm doing human stuff. And what it sounds like the direction that you're going with this is like, hold on, if AI can help us do part of it and AI can make mistakes and humans can do part of it and humans make mistakes, but if we combine the two,
is the outcome actually better, more consistent, more predictable, because AI can do the things that humans are not always very good at, like cross-referencing information of huge data sets, so that humans can do what humans are really good at to produce the right system and outcome.
Alane Boyd (08:55)
Mm-hmm.
That's exactly what I'm saying. And also thinking about where my skillset is. My skillset isn't building a 200 line spreadsheet on what was invoiced for the past year for each of those sub clients. My skillset is looking at the data and saying, this is where we can go with this corporate client. This is where I see the need and selling them on that idea. If I would have had AI back then,
and it could have done 90% of that spreadsheet and I could have gone through and said, no, this is wrong, this is wrong. I would have had that done in a matter of hours versus that was a month's worth of work for me to comb through all that data, build the spreadsheet, check the spreadsheet. Even talking out loud with some of my team members I needed to have that brainstorm on the details of that content. That timeline would have been so much shorter.
My span to talk about strategy would have been so much I would have had more capacity to think about the data versus building the data. And maybe there would be a few mistakes, but I was making those mistakes, more mistakes than the AI would have.
Micah Johnson (10:13)
I think that's the whole point of what we're trying to say here is if you do it without AI, there's mistakes. If you do it with AI, there might be mistakes. But as a human cognitive threshold here, there's only so much mental energy that we can use. That's part of the reason that it took so long.
Alane Boyd (10:34)
Mm-hmm. We're only as good as what we're in that moment. We're going to think about things inconsistently. We might have other things on our mind. That doesn't mean we're not great at our job. We're just not consistent at our job. Well, creating something like a Skill like you're saying, Micah, actually creates a consistent
perspective that a human had, but the AI can execute on over and over and over again and be its best self. And we can critique the Skill and improve that instead of always having to start from scratch.
Micah Johnson (11:06)
I totally agree. it's human knowledge going into it, the execution happens, human knowledge, judgment, experience, awareness is where our energy can really go. And we still are going to make mistakes. We're still going to make bad judgment calls. That's okay. But we've been designing systems like this forever.
Like we said earlier, since the dawn of business, we've already accounted for Everybody that has built a team, even if they're terrible at operations, still has the basic knowledge that humans are going to make mistakes. So you can use AI. It's going to be okay. Just don't think about it as it's true or false. And if it's false, then it's broken. It's just a different perspective to think about all of this.
Alane Boyd (11:30)
Yeah.
Micah Johnson (11:55)
It's why we have checklists for human One of my favorite business books of all time is the Checklist Manifesto. And it's a whole bunch of examples on how a simple list that you mark off as you're through it is one of the most effective ways to get stuff done because you don't have to forget it. You don't have to remember every single step. And you can confirm that you went through everything that you needed to to achieve the success.
Alane Boyd (12:22)
It does feel good to check something off.
Micah Johnson (12:24)
And you get the dopamine burst.
Alane Boyd (12:27)
And you know, I'm gonna take a side note on this because even if you're using something like ClickUp, Asana or Monday, and I see this common right now and I don't know why, but instead of creating a recurring task so that, you know, when you mark it off a new one gets created, they just change the date so that task never gets completed. And I'm like, my gosh, like the motivation for that task just dropped immensely because it never ends. But having that...
Just that simple check off, think really helps just that mental load of I did complete this this week and it's done. And then next week's a new task instead of it just being pushed back.
Micah Johnson (13:08)
So
this is super interesting to me, Alane, because connecting the dots on what we're talking about through this episode with that is if we were to frame this, why do we like the systems that we're building so much with AI internally for our own business? For me, the answer is because it gets us to that checkbox faster. We're able to do more.
Alane Boyd (13:27)
Hmm?
Micah Johnson (13:30)
More accurately with the help of AI, accounting for errors, accounting for mistakes. We look at it, we make the judgment calls, but then we're like, dude, I'm, already done with this. That wasn't one week that I had to stare at this task. This was one day, this was one hour and it's done, done, done. And then all of a sudden work starts to feel a lot more fun because we're looking at a tangible output that we completed in a short period of time. And it was really fun getting there. And we didn't have to do all the crap we didn't want to do.
Alane Boyd (14:00)
I love that perspective. And obviously we love getting time back and doing things faster, but also about, I didn't have to just move that data from one place to another. Automation and AI, that's prime for I've had several conversations this week, Micah, and I'm sure you have too, that a lot of what people really need is automation.
And whether or not you use AI, but automation could be alleviating so much data movement that humans are doing right now. And I guarantee you they're making mistakes at it. No one is perfect and no one really likes copying and pasting data. That is boring work. People get paid to do it because nobody else wants to do it, but that is not a fun job to do.
Micah Johnson (14:47)
Yeah, I don't know anybody that really likes doing that in a long-term basis. It just numbs your brain.
Alane Boyd (14:55)
I really think that the question, and it's whether you have a human in place or an AI in place, is not like, ooh, is AI gonna get this wrong? Because you do testing and you can figure out how to get a consistent answer, but it's really where does AI fit in the system that I built? And where does a human need to be the oversight in it?
Micah Johnson (15:12)
Yeah.
Yeah. And I mean, we covered that a lot in our last episode. So if you haven't listened to that yet, go back and listen to that episode, because we do a deep dive on this specific topic. I think that's a whole, a whole piece of it that you definitely have to account for. What I would say is the giant takeaway that I would want to convey in this episode is to not think about AI, even though AI is on our computers and it's digital, it's different.
It's not this binary system that says, true or false. has different paths that it can take, much like our own brain works. And because of that, even though it is digital, we have to think about it differently. How we incorporate it into our business is different, but it's not a whole new system design that we have to worry about. My whole point that I wanted to bring up today was that
You've already solved it. It is exactly the same systems that you build with humans. I mean, this is like maybe a great way to end the episode, Alane, is the triple check system that we put in place at our last business where we needed to account for human error so much. It wasn't a single check. It wasn't a double check. It was a triple check because our standards of operation meant no errors get to our clients
Social feeds, essentially.
Alane Boyd (16:39)
If you're wondering okay, well, what does all this mean? How do I like put this into play when you want to use AI, but you're not really sure, you know, where to do it we've had several episodes, Micah, all and Claude Cowork. because it's so incredible. But, you know, start using it for something and see where AI fits into that process. Then create a Skill for that, basically an SOP for how that can be used. Share that with your team, get them using it, get feedback. mean, we just did that.
for one of our Skills, Micah, and our team came back with some other ideas to improve it. Absolutely, like let's get it improved. Once we get it fine tuned as a Skill and with our team, then we can look at going, okay, let's automate this process more using something like n8n. That can be a now automated company process. We know where AI fits in. We know what the SOP needs to be within that AI agent to execute the same way every time.
And then we also know where the human needs to be.
Micah Johnson (17:39)
Yeah, and I love that because part of the issue with jumping right to the end result of automate everything is getting data set up correctly, getting the inputs, getting the flow. And the process that you just described, is an organic way to get to that endpoint while still getting quick wins along the way. It's genius.
Alane Boyd (18:02)
It is because this is what we see over and over and over again. When somebody jumps to automation or without AI, just when they jump to automation, they don't know the data well enough to build the automated system. And it feels exciting like, we're gonna go and automate this, but it takes so much longer because all this nuance and all these details have to be worked out in real time while building the automation. But if you...
Even though it feels slower, if you use the process that Micah and I just chatted about, it actually ends up being a whole lot faster because you know what you're building when you go to build it instead of effing around trying to figure it out.
Micah Johnson (18:41)
All right, if this episode resonates with you at all, and maybe you're one of those people that we're holding AI to that non-human standard and thinking, I'm just not gonna use it, but listening to this, you're like, maybe I should check it out. Please leave us a five-star review, leave some comments, tell us what you're gonna change. We do read all the comments that are being posted, but engaging with us.
on Spotify, on Apple Podcasts, really helps the algorithm send the show out to other leaders who need to hear this information. We would appreciate it so much, and thank you so much for listening.
