Alane Boyd (00:01)
New AI models drop every week, but the three things you need to actually get value out of AI haven't changed and probably aren't going to. Get these in place and the difference between using AI as a chat tool and using it as your coworker becomes night and day.
Micah Johnson (00:19)
So Alane, the pace of change in AI, it just makes a lot of people that we talk to just want to freeze. We've got all the new AI models. new platforms. new technology. We've got new features within those platforms. It is insane, And understandably I think if we weren't in this, we would probably be looking to other people like us to say, how do we get started in this?
Alane Boyd (00:44)
Yeah, I really started to think about this a lot over the last couple of weeks because the pace of AI updates coming out is, so much faster even than it was a year ago. There are things that are changing and platforms are constantly just being the next best. But at the core of it, we're really seeing there's some things that haven't changed. It's been the same thing for 15, 20 years, maybe of all time doing business.
and it definitely relates to how you use your AI.
Micah Johnson (01:14)
Yeah, it's pretty cool that we narrowed this down to three things in this episode that we're going to talk about. these three things are going to help you in business period, and it's going to help supercharge your AI usage. But it's just really interesting to me to think from a systems perspective, holy crap, the fundamentals are here. Here are three things. Start building these today. And it doesn't matter what platform, feature, model you use.
Alane Boyd (01:22)
Mm-hmm.
Mm-hmm.
Micah Johnson (01:40)
because all of them these three foundational elements.
Alane Boyd (01:45)
Yeah, I was thinking, this week or last week, time's getting away from me is kind of that freeze that you feel when OpenAI just announced, you know, that they're going to be sunsetting the custom GPTs and moving their workspace agents that is scary. And if you've invested a lot in custom GPTs, you know, undoing that can feel like a lot of work. now I'm like, well, it really comes down to the three things that were going
talk about today because those things are transferable no matter what if you switch AI platforms.
Micah Johnson (02:16)
Yeah, very true.
Alane Boyd (02:17)
So should we cut to the chase and talk about the first one?
Micah Johnson (02:20)
Let's
stop teasing people about what these three things are and let's jump in at number one, which is clean, organized data and knowledge in a reusable place. We specifically use the term reusable place, Alane, and that's not a technical term I would say that most people come up with. So why don't we talk a little bit about why we chose to describe it that way?
Alane Boyd (02:22)
Thank
Mm-hmm.
you
Well, I think the first is the context that we need to give AI. You and I talk about context, even when we were talking about prompting, you needed to give AI some now that's still relevant and how we give that, we don't want to keep reinventing that context. You don't want to have to every time say the same thing over and over again to AI. And so we have distilled it down to one place
really, that has that information.
Micah Johnson (03:14)
Yeah, and this trend, this is a foundational element of business and operations that have always been there. But it's making the rounds more so now because people are doing things like, well, let's make a knowledge brain and then attach that to Claude Cowork. And we've talked about that on this podcast. We use that and it's brilliant. And it's actually
forced us, and is a strong word, greatly encouraged us to get really good about not reinventing the wheel, about not rewriting things, about not having version issues all over the place and putting it in a reusable place that is the source of truth, this is the curated thing, and that's it. We draw the line right there.
Alane Boyd (03:50)
Mm-hmm.
So one of our reusable places is We've been using Dropbox even in our previous company. So not just in the last couple of years, we've been using it for a long time. And even if we're building a RAG database that an agent has access to, you know, we can tie that into our Dropbox. That is a reusable data centered place for our stuff. And now...
we can build a RAG database off of that, or we can also tie it in with Cowork who has access to it and has that information to go back to every time.
Micah Johnson (04:34)
Yeah, and this is, again, from a foundation point of view, it doesn't matter if you're still using chat versus Cowork versus Co-Pilot versus OpenAI. Create that reusable place, create that central source of truth, create that knowledge, whatever term you want to call it, create that and then use that and curate that and
take the effort to keep that up to date and don't let it fade. It pays back in spades with AI and it helps you across your entire business period. Alane, there's one thing that I do want to point out with this it's alongside of this that people go, oh, well, I have 900 files across.
Alane Boyd (05:03)
Mm-hmm.
Micah Johnson (05:20)
one folder, it's everything we've ever done, it's every version of our SOP, it's every version of a email that we've sent out as a welcome message. Why can't I just give that to AI and have it figure everything out for us?
Alane Boyd (05:39)
Yeah, talk about context a lot, like I mentioned, and that's way too much.
Micah Johnson (05:43)
I've never said context
more in my life than like the past three months.
Alane Boyd (05:48)
But that's way too much information. Like the AI just cannot comb through all that without giving up basically. It's way too much. So just like you wouldn't tell an employee that's brand new, hey, yeah, that's in Dropbox and it's got 900 files and they have to dig through it. That would take them hours digging through and trying to find it. So clean data has...
Micah Johnson (05:55)
It's gonna guess.
Alane Boyd (06:12)
always been important, but even more so now because you want the AI to be working efficiently and getting the job done.
Micah Johnson (06:18)
Yeah, perfect explanation. clean data, keep it organized, keep it structured, curate it is my new, I'm going to say curate more than context. That's my new self-challenge it's going to help your business. It's going to help you work with AI. Number two, documented SOPs. So nobody is guessing. We were originally talking about this just in the context of AI, Alane, but I...
Alane Boyd (06:27)
Ooh.
Yeah.
Micah Johnson (06:45)
change the words at the last second. And I'm like, wait a second. It's not just so AI isn't guessing on this. It's so that nobody is guessing, nobody on your team, not yourself. We just got through a mastermind session earlier today, Alane, and I admitted to the whole group. said, hey, I'm Micah. And although I love processes and procedures, I'm also addicted to making new ones.
you know this better than anybody, Alane, I love writing a good process. I don't love following a good process.
Alane Boyd (07:11)
God no.
I was thinking about that yesterday, because Micah, you didn't follow the process. And I'm like, he just loves building a process. He doesn't want to follow it. Drives me crazy. I could hear the listeners groan Micah when you said documented SOPs. Yeah, well, that isn't the most exciting part of a business. We started off even helping.
Micah Johnson (07:29)
I could hear them relate. They were like, yeah, that's me too.
Alane Boyd (07:41)
Businesses document their SOPs, but you need them unless you want to be doing everything every single day or reinventing the wheel. But an agent or a Skill in Cowork, whatever it is, at the end of the day, they have to have an SOP to work from so they know what they're doing and what the outcome is and how you want it to look.
Micah Johnson (07:59)
Yes.
The best part now is you can leverage AI to help you write the SOPs, to form it in a way so that you look at it and go, that is the outcome that I want to get and that's the way that I want to do it. Have AI help you write those that used to be super time consuming and tedious. Now you have AI help you write those. You organize it in what was number one, ⁓ clean knowledge, data, reusable place. And you put all your SOPs in there and
Now you have the AI doing the same thing in the same way every time. If you're using chat, you're going to have to copy and paste. If you're using Cowork, you're going to be able to give it access to that folder system and it can pull out the SOP. You can even create an entire Skill that is essentially an SOP for AI. But if you don't sit down, whether it's yourself or with your team or your leadership team and say, how are we going to do this?
what's included in this, what's the output that we're looking for and define and standardize it, then everybody's going to do it a different way and you have siloed implementation. And again, AI or not, this is a major best practice just to help scale and grow your business. And just by going through this exercise is going to change your world.
Alane Boyd (09:16)
So, I wanna be clear, Micah, because it feels like we're saying like an SOP for your business, which we are saying, but the agent or your Skill in Cowork is operating from an SOP. So whether you're using it for your people or for your AI, those SOPs are needed. And that's essentially, even if we're just talking about Cowork, a Skill is just an SOP. So having those documented,
Micah Johnson (09:39)
Yeah, good point.
Alane Boyd (09:42)
feeds into your AI so it can do the job repeatable. We've got one reusable and two is repeatable.
Micah Johnson (09:45)
Well, and just like if you ask.
Yes. And just like if you ask somebody to do something by nature, a person is going to just guess at how, you just say, Hey, here's, here's what I want done. And that's all you say. They're going to guess based on their knowledge, their experience. And if they don't have any of that, probably going to be pretty bad judgment calls and pretty bad guesses. That's where AI is at every day. It's a new day. Every chat is a new chat.
It doesn't have everything like you have in your head that can look at a problem and go, this is how I do it. You have to spell it out. You have to give that to the AI otherwise it's going to guess. And when AI guesses, that's where it gets really nasty and bad. this is a broader topic. Sometimes it, it's, you know, interesting to see how it guesses, but when you're trying to get work done,
when you're trying to get repeatable outcomes and when you're trying to scale with AI, you don't want guessing.
Alane Boyd (10:45)
This is really a great opportunity too for people to look at their processes when they're going, hey, we want to automate this with AI to actually get a company process. You know, we've been really talking about consistency lately because it is such a big component of automating with AI. But this is where, what do you want that SOP to be in coming up with that company process instead of the Alane way of doing things and the Micah way of doing things.
Micah Johnson (11:12)
Yes. All right, Alane, I'm going to ask the listeners for a quick favor before we get onto the third one. If there's an SOP or a point that we just made that helps you think of a process that's in your business, that's living entirely in someone's head, leave us a comment. We'd love to hear where people are getting stuck. We've been there. I'm sure other listeners would benefit from it too. It also helps us to make sure that what we're talking about on these episodes,
are landing for everybody. Our listenership is growing so quickly right now and it's super, super exciting to see. But the more that you can comment, the more that you can like what we're doing, not only is it good feedback for us and it gives us warm fuzzies, it also tells the algorithms that, you should probably share this with other like-minded people and then other founders, agency owners, business owners get to hear this valuable stuff too. Would really appreciate it. Thank you.
Alane Boyd (12:11)
All right, and number three, Micah, is another one of these things that we've been talking about for so long and is the most relevant of changing to an automated fashion with AI. And that is working with platforms that have APIs, webhooks, or now the newer version of those two things, MCPs.
Micah Johnson (12:30)
Yes, before we dive into this one, I mean, this one makes my stomach hurt because there's so many conversations before AI where we're just trying to build automation. We're trying to build scalable systems. We're trying to help companies do that kind of stuff before AI. And then we'd run into a great company and they'd go, well, we're working with this platform that doesn't have web hooks, doesn't have an API, doesn't have any integrations, can't connect it, can't connect to Zapier, can't do this. And their only option.
is to throw people at it and throw money for the people that are doing these monotonous tasks like logging into an application, going to the same spot every day, pressing the same filter buttons, and then pressing an export button, and then getting the CSV file that was exported, and then getting that into wherever it needs to go or manually processing that from there. That hurt. But now with AI,
Alane Boyd (13:26)
you
Micah Johnson (13:28)
Right? It's kind of like saying, ⁓ well, all these other companies can do things in five seconds, but sorry, everybody, we can't because we have a platform that can't connect to anything.
Alane Boyd (13:34)
Mm-hmm.
I feel bad for the software companies that don't have those things because to stay relevant, you and I do not pick a software now that does not have an API web hook or MCP and even more so an MCP now because of what we can build within seconds versus months go in the other route. And if you are using a software that does not talk to any other software, which is what
an API, webhook, or MCP does, it gives that ability for software platforms or AI to talk to it. Then you're using something that just is not functional. You have to be looking at an alternative. And that to us is crucial in using tools now. There has to be a way for it to conversate with other platforms. And we love when it can automate in Cowork. So an MCP server is awesome.
Micah Johnson (14:31)
I
We've seen this transition before. I think it's even bigger than anything we've seen in the past. A transition like this before was a software that was only in DOS and could not support the graphical interface of Windows. There's a big functional difference between those two. Then it was a desktop application with zero internet connection and
a online SaaS product. that introduced the whole SaaS industry of the past 20 something years. But now have software, if you're building software, if you're a software provider, you have to have these connections. You have to have this integration because everybody is going to be using AI. And if your software can't handle it or can't support it,
Alane Boyd (15:05)
Mm-hmm.
Micah Johnson (15:24)
then you're forcing them to go old school, your users, and it's just, it's not gonna gel. Now the good news to all of this, Alane, is that what I'm seeing in the trends that are happening is that there is a whole new world of software products coming out that are AI first, MCP first. And when I say AI first, I'm not just talking about generative AI that's like sitting there on top of what they already did. This is like,
Alane Boyd (15:27)
Mm-hmm
Hmm.
Mm-hmm.
Micah Johnson (15:51)
Reimagining CRMs with AI at the center. Reimagining design software with AI at the center. And those have the APIs, the webhooks, the MCPs, everything that we need to connect these AI platforms into it. So hope is there. The downside is if you're still on an older version of something that doesn't have it, migration is a term you're going to be saying a lot in the next three months.
Alane Boyd (16:14)
Mm-hmm.
You're gonna be migrating.
One thing that I wanna be clear on is it's almost the older companies that haven't kept up, like a software platform that's been around for 20, 30 years is having a harder time catching up than a startup a year old. I mean, Anthropic's only like three years old that owns Claude.
So it's the newer companies that are building technology the way that it needs to be instead of these old companies trying to catch up. So we're not saying don't go with a startup or a newer software company. It's looking at who's adapting at the rate that they need to with these APIs, webhooks and MCPs.
Micah Johnson (16:58)
Yeah, I totally agree. just coming from, how I look at the world, I think this is actually one of the biggest opportunities we've ever seen. There's all this old software. it can be slow to, catch up. It's very hard to we have this giant code base from 15 years ago. We've got to refactor everything. We've got an add an API. Then we've got to figure out MCPs.
then we got to figure out webhooks. Those are real costs to software companies. The opportunity is, well, hell, I'm in a specific industry. I see a specific problem. I see no software with APIs. Let me talk to my good buddy, Claude Code, and see if I can draft up a prototype. Now, I'm not saying go production with Claude Code and vibe coding everything, but
Almost anybody can take an idea to execution now, even for this. And this is getting way off topic, you know, there's so much like doom and gloom around, what are we going to do when AI is our new zookeepers versus there's actually silver linings on this. There are opportunities that we can look at. We were talking about it even earlier today that it's like companies being able to build their own internal tools actually is going to open up a huge wealth of
hiring, I think, for people that are going to be able to maintain those internal tools because the same people that build them are not always going to be the same people who want to maintain them. I fall in that category. It's just like processes.
Alane Boyd (18:26)
That's true. So I think like what we're saying here for all three of these things is that they haven't changed. We've been talking about these three things for over a decade, even before we had this business. And so now looking at what is your business documenting? What software platforms are you using? And if you have clean data
we just talked about like a Dropbox or a file storage, but there's also a whole other data set for like Airtable or SupaBase is starting to look at those three things at your company. And if you are successfully doing those three things, then it doesn't matter so much what AI you use or how things change, because those three things are going to be at the core of your company on how AI uses it.
Micah Johnson (19:12)
Couldn't have said it better myself, Alane.
