Loading summary
Transcript276 lines
- [00:00]
Dan Cumberland
AI is going to take jobs and it's taking jobs. AI is going to change all of our work. If there's one thing that I want every business leader to do differently, it's to my guest.
- [00:09]
Nathan Berry
Dan Cumberland is a serial founder with multiple successful exits who now builds human centered AI systems for creators and entrepreneurs.
- [00:17]
Dan Cumberland
Here's the trick. I call this the turnaround. Go back to your chat and then just say print a prompt. That would allow me to take this whole conversation to another AI chat and start over.
- [00:27]
Nathan Berry
His work focuses on helping high earning creators and companies implement workflows that make it so you can actually enjoy your work again.
- [00:35]
Dan Cumberland
What if we can zoom out and see like what is the real work that only I can do? I think we're underselling ourselves if AI is that big of a threat.
- [00:43]
Nathan Berry
In this episode we show you exactly where you're going wrong. How to set up better systems to get precise, accurate output that's going to change the way you operate in your business.
- [00:53]
Dan Cumberland
It is like rocket fuel and nobody knows about it. Teach Claude how to.
- [00:57]
Nathan Berry
Okay, this is a key tip. Dan, when we were talking earlier, you said this line that really stood out to me and you said with AI, I'm able to be the creator that I've always wanted to be. Tell me about that. What does that mean to you?
- [01:11]
Dan Cumberland
So I've been creating on the Internet since probably similar, similar starting time. Like early world domination summits. I was there and you know, just have always been like, it's so important to me. But consistency is hard and where I, where that takes me is like there's just friction in the process. And so when I, what, what I love most about is for content creators and for knowledge workers in general, it allows us to take the friction out of the process and, and get faster into the work that matters most to us. And that's really like kind of my passion behind all of my, all of my work is like getting people into their zone of genius and doing more of that thing. So for me personally that's what, what, what it looks like. And most specifically with Kit, just to describe this workflow, I create content usually the top of my or starting point of my, my content Flywheel is, is LinkedIn. I'm posting on LinkedIn regularly. I take some of those posts and I turn them into broadcasting kit. So first I'll take that post, I'll maybe beef it up a little bit. I'll use my AI to, to help help in that process, really making sure it's, you know, still on Brand still my voice but you know, just again making it go faster, easier. But then I have to get into kit and, and click the buttons, copy, paste, do the formatting, you know, make the button so it's linked to the thing that I'm linking to and all that stuff and it's just, it doesn't take that long, it's like 15 minutes. But like I find sometimes that like I don't, I don't send my newsletter cause I don't take that 15 minutes cuz it feels like a hard 15 minutes. So what I've done is using a browser automation plugin with Claude. Claude taught itself how to, how to go into my kit account, do all the formatting. It says an AB test in my subject line, formats my entire email in exactly the way that I want it to be. Does the button in my brand color and then goes to the next page, the scheduling page, sets the preview text and like an internal note about what that subject that, that email, that broadcast is all for me. And then I just go into my kit account and it's just ready, I can just.
- [03:12]
Nathan Berry
You just proof it and it's end. Yeah, that's such a great practical example. And I find that in AI people are throwing out all of these high level, like it changes everything but then we're like well what do I actually do? And so it's all theoretical, like all theory, very little practice. One thing I was so excited to have you on is every time I talk to you it's all like practical implement like tips to do right now. So what we're gonna cover in the episode, you've got three different tips that people should implement right away. You've got five practical examples that we're gonna run through and get into in a lot of details. But the thing that stood out to me most when you talk about being the creator that I want to be, the type of creator that I want to be and AI allowing that. It's about reducing friction, isn't it?
- [03:54]
Dan Cumberland
Yes.
- [03:55]
Nathan Berry
How do you, how do you find the friction?
- [03:57]
Dan Cumberland
How do you go about reducing friction? Yes, I love it. Yeah. So we all. The promise of AI usually has to do with time. Like you know, you can get so much more done. You can have all your swarm of agents go do your whole workday for you so you don't even have to work and all of, all of that, you know, hyperbole. When I work with clients I like to pay attention to where they are, not sometimes about literal time. Like where, where can, where can we Buy back your time. But sometimes what's more meaningful than literal time is felt time. And that's marked for me by friction. Like what are the things in your day that like, I would rather gouge my eyes out than do this task. It might only take me 15, 15 minutes maybe. You know, for me, you know, it's, it's loading those things up, clicking a bunch of, clicking a bunch of stuff. Other people might be accounting things, but like what are the parts of your workday where you're experiencing that friction? Cuz that friction I think has downstream effects, leads to burnout, leads to, you know, oh, just extra stress. How can we streamline those things and get you closer to the work that only you can do? Which usually isn't that friction. Usually it's the stuff on the other side of it. So I think it's a really important indicator.
- [04:57]
Nathan Berry
I love that so much. So diving into the practical examples, what's another example of how you've implemented AI for a client?
- [05:04]
Dan Cumberland
Yeah, so let's, let's. I'm going to just go through some of these because there's this, there's this chasm right from where we are and all the promise and possibility and the best way to cross that chasm is by seeing how other people have crossed it. Because, because it gives you just real practical tactical ideas even if you're not doing exactly what they're doing. So first one client that I had is a podcaster I've worked with and ended up working with a lot of, a lot of podcasters, a lot of content creators in particular. But she had ST Rapaport. She's a fantastic mind brain coach. Specifically she works with a lot of, a lot of ADHD entrepreneurs, but not, not exclusively. And she has a solo podcast. She was already scripting her episodes. She came to me and she was like, I, I, my top of funnel is, is TikTok. I right now I'm like, as far as like how I can use AI, I think like there's something here that might help. Right now I'm chopping up my podcast and putting it on TikTok, but it's like, it takes a lot of time. Could AI help with that? So we, at first we dug in, we're like, okay, like can we use some tools to like help with, you know, opus clip and like those kinds of things that like take long form content, make short form content. I built a product, I don't know if you remember that did exactly this back before, before opusclip when I was running A venture. So I played that game before. But one of the things that I know about TikTok is that clips can do well, but content created for TikTok usually does better. She already has this great, this, these great, this great ip, these great thoughts that she's already articulated. How can we package that, get it back to her in a format that's better for her, ready for her on TikTok. Most of her TikToks are very short, like 20 seconds or less. Just like really, really punchy stuff. So what we did was we took her. Her content is already, already like every episode is already in a Google Doc, in Google Drive, but so, so clean and so, so easy. Then we developed a series of prompts that would take each one of those episodes that's already her words and then format them with different hooks, different formats for TikTok into scripts. So scripts fed back to her from her words in her own words, in new formats. So then she could pull up her script on her phone anytime she wants to record and just record each one of her episodes then turns into 12 TikTok videos. But here's the thing, she had 500 episodes in the backlog.
- [07:26]
Nathan Berry
Oh, wow.
- [07:27]
Dan Cumberland
So we built. So the first thing we did so you could, you could take that same for everyone listening, take a very similar process. Think about what's your first, the top of your content flywheel and how, what are the other places that you want to put that content and how can you intentionally build a prompt? And we can talk about this a little bit more later about context and that kind of thing. Context engineering. Teach your AI to, to take that and put it exactly into the exact format that you need. So that's a takeaway that everyone should be taking from this. But then in some cases you can automate this process. And it was so clean for her because she was already doing this. And when she wanted to record her TikToks, she, you know, would, would pull up a spreadsheet, have her, have her, her scripts there. So we ran this automation, we built it in, in Zapier with just a couple prompts. You could do it without Zapier just in Claude or Chat GPT, but Zapier can just automate this kind of stuff. And it just powered through all this spreadsheet, you know, 500 times 12. I don't know the math. It's a lot. Yeah. And then so we built all that out and then every time she, you know, now it just watches her folder every time she drops a new, a new doc Trans, like, script for her podcast into that folder. It. It prints another 12, 12 scripts in her spreadsheet. So, like, literally, she has more content than she'll ever be able to get to because there's so, so many of them. So that's. That's workflow number one. And. And again, think about where you're creating and then how can you take that? Use AI in a really intentional and, like, clean workflow. Workflow design is super important in order to repackage it into the next piece of your content. Flywheel.
- [09:06]
Nathan Berry
So anyone who listens to this show a lot knows that I love the theory of Constraints, which is all about finding the bottleneck and obsessively fixing that bottleneck. And so for most people, if they're saying, okay, I want to grow on social, I need. My bottleneck is ideas of what to create for each of these Instagram or TikTok reels. And so in this case, you said, okay, we took the bottleneck and we
- [09:27]
Dan Cumberland
just, like, blew it up.
- [09:29]
Nathan Berry
That is so far from the bottleneck now. You know, now the bottleneck might be editing or posting or something else. Right. But she can sit down and she could actually record 10, 15, 20 of these videos, do it in different backdrops, different, you know, a bunch of those things, and churn it out and have endless ideas.
- [09:47]
Dan Cumberland
Yes.
- [09:47]
Nathan Berry
What. As you look at her workflow, what's something else that would come up? Yeah, like, we've now moved the bottleneck.
- [09:54]
Dan Cumberland
Yeah. What's next?
- [09:54]
Nathan Berry
What was the next bottleneck? What came from there?
- [09:57]
Dan Cumberland
Yeah, yeah. So the. The two different places I look are one, like her podcast creation process or optimization. So how is she looking at what's performing well? How is she getting data from the platforms and then using that to feed the system? This would be a more as far as like, an automation, a much more complex thing to build. But you could do it manually by just looking and seeing, like, okay, these particular formats are what do best for my, you know, for me. And so maybe we need to adjust the prompts so that we have more that are like that or like, some things like that. That's where I would look at optimization and where the next. The next layer could be.
- [10:32]
Nathan Berry
So something that those of us who create content at scale usually don't do is analyze it.
- [10:38]
Dan Cumberland
Well, yeah.
- [10:38]
Nathan Berry
Or, you know, categorize and then analyze. And so when we were talking earlier, you were asking like, hey, is there something if you were to, you know, wave a magic wand or like, what would you want AI to do for you in this Example what I would want is, I would want automated analysis of is this working? So congrats. You can now create 10 times the amount of content with, you know, the same amount of effort. Yeah, but like, is it worthwhile?
- [11:03]
Dan Cumberland
Yes.
- [11:03]
Nathan Berry
And I feel like that's a problem that AI could solve. So how would you go about analyzing the social stats? Maybe you do it for your LinkedIn post already to see these categories are working. Do more of this.
- [11:15]
Dan Cumberland
Oh yeah, man. So I'm not an expert on TikTok, so I don't know, I don't know what the constraints of that platform are. But like the first question is where do you, where do you have the data and how do you get to it? We talked a little bit about how I'm using kit, long time kit user. Love the platform and love how you guys are building it for AI to be AI enabled. So one simple. This is an aside I'll come back to. This though is like for my broadcast, I want to do this very thing for myself. How do I make my broadcast better for my particular audience? So I connected to the API and had cloud pull in all of my broadcasts and then run a analysis of them to optimize my headlines. What headlines perform best, what words, what my send times, what sometimes lead to the, to the, to the most opens and then also like what call to actions do best. So now when I'm creating broadcasts, like it's all optimized based on that. You don't have to do that with cloud. Like you could export your content. I mean you don't have to do the whole API thing. It's just, it's fun. So I did that way. Um, but I would look, look to something like that with LinkedIn. I do something very similar. Um, LinkedIn's tricky because they really, they, they really have like high walls around their garden.
- [12:26]
Nathan Berry
Right.
- [12:26]
Dan Cumberland
And they don't like you jumping over them or going under them. Um, but so you can, you can tie into some tools that, or go use, use platforms that already do some analytics. Um, fun fact about all of those platforms is that they are all based not in the United States. Because if you create a LinkedIn analytics platform, you will get a cease and desist from LinkedIn. I know because I created a plugin that does this for me. So it looks at my, it looks at my content and then just tells me what's well. And I was like, oh, maybe I should turn this into something and sell it to someone. But then got connected to just as I was starting to explore that Someone else who had a platform that just shut down. I was like, hey, why did you shut this down? It seems like such a great idea. I was like, because we got the cc, you got this letter. And so I found my way around that. I shouldn't be talking about it publicly.
- [13:17]
Nathan Berry
I wonder about speculating some ideas of browser automations or browser automations or like,
- [13:24]
Dan Cumberland
you know, use the tools that, that do exist, Taplio and, and Shield analytics, those kinds of things. But the idea is like, you just gotta figure out how to get that data and then use AI to find the patterns in that data. And I like to, you know, cross check that with, you know, what's, what's doing best and then look at my drafts and like, what is similar, you know, what are the ideas that I have not yet used? Or as I'm gonna write new content, let's look at what has performed well and how can I make my new content like that other content. You can also do that based on other people's content, see who's growing well and use, use AI to help you optimize your content for like the growth patterns that are happening. And that's helped me a lot. Had like about a month ago, I had like three posts go viral in one week. Did like almost 400,000 impressions, which I didn't know was possible on LinkedIn. Maybe some people watching are like, you know, Chris Donnelly style and just crushing it. But I'm, you know, grew my, my, my, my LinkedIn followers by like 4,000, which is a lot for me. Um, so like it does work, doesn't always work, but it moves you in the right direction, prepares you to go, to go viral. So to just zoom out again to system design. You want to think about where, where are the numbers and then how can you feed those numbers back into the system to let it help you optimize your content.
- [14:40]
Nathan Berry
I love that. Okay, I want to go to some quick tips. Yeah, you had, you had a couple, like, what are the three things that you think that you would most want people to implement or to know about what they can do with AI?
- [14:52]
Dan Cumberland
Yeah, yeah. So the first thing I always think about is, is context engineering. So and this is like the one thing that, like, if there's one thing that I want every business leader to do differently, it's to not trust, not trust AI, but, but more, more than that, it's to, is to not, not rely on AI's knowledge of you or your business to accomplish your goals. So we talk a lot about prompt engineering and maybe this is, you know, is, is maybe not as, as hypey as it was, you know, maybe six, six months a year ago, because the models are getting more sophisticated so they're understanding our intent more. But what, where we are now I think is really in the age of, of context, the era of context engineering, to think really intentionally about how you're giving the AI everything that it needs to know in order to do the thing that you want it to do there. With every one of my clients that I work with. And so I work with both in a group setting with, with, with coaches, consultants and solopreneurs, people doing 5 million and less, where we help them apply AI in their own business with the promise of saving them 5 to 10 hours a week by the, by the work with the workflows that we build over that 12 week program or I build it for them, which is a big promise. But we, so in the very thing, first thing we start with is that question that you mentioned earlier. Like if you wave your magic wand and point at something, what would you, what would you point about? So both with them as well as where, where I'm working, we're in a done for you fashion. With bigger companies. Five to 50 million are, are usually the size of the companies where regardless, it always starts with these five. These five core, what I call them is context documents. And a context document is just a document that says a lot about whatever it is that you need. Um, you could think of it very similar to prompt, but it often doesn't have a specific outcome. It's used as a reference into your custom GPTs, your gem, your Gemini gems, your cloud projects, et cetera. And so those five are rich, robust context about the company or the person. So what, what the brand is about, mission mission vision, vision values, their. Why all of that? The next is the, the, their icp, their, their customer. Who are they trying to reach? What are their, you know, the psychographics, the demographics. The next is a style sheet that teaches the AI exactly how to, to create in that brand's voice. So really focused on language and the linguistics, even as the specific platform if we're creating for something very specifically. Um, so like you know, we write differently on LinkedIn than we. Then we script a podcast or something along those lines. Um, and then the next one is their product ladder. What do they sell? Who do they sell it to, at what price points? And then the final one is stories and, and anything else that, well, this will depend on the specific output that we're after. But that Gives more depth to everything that we just shared before. And so the biggest to the mistake is just instead of doing that work and really giving AI everything that it needs in order to produce high quality output, we just say, my AI knows me really well. I gave it a PDF of my website and my LinkedIn profile, and that's not bad, but it's just sloppy. And so the invitation is, instead, have a conversation with AI to help you produce the kind of content. So AI can help you make that, but don't just throw your content in there and say, I've got, I've got all the context I need. So use AI to context engineer the things that you need in order to have a higher quality output.
- [18:24]
Nathan Berry
Something else that I think I've heard you touch on in the past is a mistake, really that ties into this is relying on the memory of that individual chat or that tool.
- [18:34]
Dan Cumberland
Yeah.
- [18:34]
Nathan Berry
And so it's like, you know, I've been having, going back to the same project for years. Probably not years, right?
- [18:40]
Dan Cumberland
Because, you know, feels like years, but
- [18:42]
Nathan Berry
then, you know, for months.
- [18:45]
Dan Cumberland
Exactly.
- [18:46]
Nathan Berry
And I keep having these same conversations and it knows me so well, but then something comes up, like something that you and I both believe is Claude is remarkably better than chat right now. And six months ago, I don't know that that was the case. Yeah. And so if you have this perfect memory in ChatGPT and then you come along, you're like, claude's way better. I'm like, oh, but I can't move out of this. So good, because my context is not portable.
- [19:09]
Dan Cumberland
Yes.
- [19:09]
Nathan Berry
And what you're saying is, get into these documents now, your context is portable. And someone comes along like, we're raving about Claude, and someone's like, actually, Gemini 3 is pretty great.
- [19:18]
Dan Cumberland
And we're like, all right, well, let
- [19:20]
Nathan Berry
me give it a try. My context is portable. I have my five documents, they're loaded in. Let me try the same prompt in both and go, oh, well, I do like Gemini for this thing.
- [19:29]
Dan Cumberland
Exactly. And so then you can rely on all of them for their strengths. And so by doing that, you can, you know, like you said, you, you become less, you know, I call, like AI agnostic. You become more AI agnostic. Less. Less locked into, you know, a subscription especially. And that's, you know, that's what's going on behind the scenes. You know, I think that like, ChatGPT and ChatGPT and only fans have the same model. Right. It's subscription based. And so, like, how can you, you know, how can we give you the opportunity to, to take your content elsewhere. But even in your specific example, which I love that you brought that up, like we all have these like long chats we just keep going back to and like, oh yeah, I got to go do that one thing. And like I haven't, like, I don't know, I haven't built a project for this or whatever. So I just go back to that same one and it just grows longer and longer. And over time there's a couple things that happen. One is context windows, which is like basically how much AI can hold in his memory at one given time. Eventually as a conversation gets longer, you could think of it like there's a window, your chat is starting at the bottom and eventually it fills up off the top and then it drops off the window and so it'll start forgetting things. And this happens a lot less now than it did like a year or two ago because the windows have gotten bigger and they're going to continue to get bigger. But so that's one thing that can happen, especially if it's really long, especially there's a lot of documents in there, is that you'll, it'll start performing less. But then also especially if it's a conversation based, where you're covering a lot of different things and not just one specific output is going to get a little squirrely on you. And studies have been done about this that the, the less linear a conversation is, the less predictable the output will be from that chat. So here's the trick. I call this the turnaround. Go back to your chat listeners, anyone who's, who's listening right now. And you're like, oh yeah, I've got that one chat. Go back to it and then just say print a prompt that would allow me to take this whole conversation to another AI chat and start over and then tell it to especially I like to say print it in a code block. A code block formats it. So there's usually a copy button at the top. So it's really easy to copy the whole long thing. I also say like to put it in markdown, which is a protocol for formatting that's just cleaner for AI. So that's just a really great trick to be able to take any context and move it to another chat or turn it into a custom GPT or something like that. And like we're talking about this context engineering. Using AI to, to using AI to use AI. Using AI to shape how you use AI is a great starting place for building these kind of context documents.
- [21:54]
Nathan Berry
Yeah, and even saying, I've talked to lots of people who say, you know, I want to create a document like this. Interview me.
- [22:01]
Dan Cumberland
Yeah, yeah.
- [22:02]
Nathan Berry
Until I can create this.
- [22:03]
Dan Cumberland
Right. That's exactly. If you're using.
- [22:05]
Nathan Berry
Whether you're using the voice inside of Chat GPT, like, directly, or you're using whisper flow so that you can talk, it can be like this big thing of, like, I now have to make five documents. Dan said, it's like, it's actually fairly straightforward. You can make them sequentially and go from there.
- [22:22]
Dan Cumberland
Yes, yes.
- [22:23]
Nathan Berry
Two random things that made me think of one. It's wild that, like, they invented artificial intelligence and yet they're like, desperate to keep a 20amonth subscription because we're like, I don't know, this is the greatest thing the world has ever seen in Chat GPT. But honestly, like, Claude's a little better and so I'm taking my 20 over there. Just from like a business perspective. Moats, all of that. You're like, that's wild. That didn't succeed in creating a moat as they spend hundreds of billions of
- [22:51]
Dan Cumberland
dollars and they're losing money on those subscriptions. Right. And it is, it's a wild, terrifying thing. When you look at how these business models are working right now and how our economy is propped up on the success of AI, It's. That's a. That's for another episode.
- [23:04]
Nathan Berry
But we're saying practical, not existential crisis.
- [23:08]
Dan Cumberland
They're sweating.
- [23:09]
Nathan Berry
But there is that. You're like, oh, I just invest in the s and P500 and all that. And you're like, yeah, what companies make up that?
- [23:15]
Dan Cumberland
Exactly. There is like, the balanced S and P, where it's, where it's balanced, not based on. On market cap. That might be a better investment right now. Not investment advice.
- [23:24]
Nathan Berry
Who knows? But the other thing that I'm thinking about is how you share information between these windows.
- [23:32]
Dan Cumberland
Yeah.
- [23:33]
Nathan Berry
And something that I like to do is I will test whether or not my document created the same output.
- [23:39]
Dan Cumberland
Yes.
- [23:39]
Nathan Berry
So if you say, hey, print, like, print a prompt, all the context that would be necessary to start over again, then you can say, okay, I'm in. Still in Chat GPT. Yeah. New window.
- [23:49]
Dan Cumberland
Yes.
- [23:50]
Nathan Berry
New context.
- [23:51]
Dan Cumberland
Yes.
- [23:52]
Nathan Berry
Upload that as a project and then try the same prompt in your old conversation and that one, See if it's different.
- [23:58]
Dan Cumberland
Yes.
- [23:58]
Nathan Berry
Bring your content over. What I would do then is like, fresh cloud window.
- [24:03]
Dan Cumberland
Yeah. Try it there, try it there.
- [24:05]
Nathan Berry
No context.
- [24:06]
Dan Cumberland
Yes.
- [24:07]
Nathan Berry
Fresh cloud project.
- [24:08]
Dan Cumberland
Yes.
- [24:08]
Nathan Berry
Add in the context and try it and just look at the Differences and see what you notice.
- [24:12]
Dan Cumberland
A hundred percent. A hundred percent. No, I think that that's, that's a great, a great workflow just to like, trust, like, which, which, I mean they all have different personalities and that's like, you know, I think, I know you think about this with, with Kit, like you choose your customers and then you're. It's like Marshall McLuhan and like, you know, you shape the media. You, you, you. What is the quote? You shape the media and the media shapes you. Right. I think the same is true in product. Like you choose your customers and then your custom. And, and so I think about that a lot with like, who is, who's chatgpt for? And who's, who's Claude for? And like cloud is built as kind of a developer tool and it's becoming, you know, more and more, you know, user. User friendly. But they're also leading the way on integrations. MCP servers, which are a lot like APIs but built, you know, specifically for, for AI to, to access and have like levers and buttons they can push inside other software, which is like mind blowing and so cool. But like cloud built that technology and ChatGPT, you know, they recently hired the head of, of apps from Met, have a shopping feed. Like, there's, I feel like there's like this divergence happening and like, I don't think that ChatGPT is going to go away, but I do think that Claude, in my opinion, is winning the race as far as like a more functional business tool.
- [25:21]
Nathan Berry
Like pro tools versus consumer.
- [25:23]
Dan Cumberland
Yes, exactly. Exactly.
- [25:25]
Nathan Berry
Yeah, that makes sense. Yeah. The other thing that I feel like you can tell a lot about a person based on you ask, you know, hey, do you use a different chat for each thing or. The number of people that I talked to are like, no, I just. AI is one chat to me and I just continually. So they'll ask you a research thing and then like write this for me and all of those and you're like, huh, okay. You and I are.
- [25:45]
Dan Cumberland
We're different. We're different. We're different.
- [25:48]
Nathan Berry
You're talking about context. Yeah, we're kind of jumping around a little bit. But if we like litter in these, these key tips, one that I think is really important is that inside of Claude, you can pull in your Google Doc, your, you know, Google Doc or your sheet, and it will update that context in real time.
- [26:09]
Dan Cumberland
Yes.
- [26:10]
Nathan Berry
And so I, I think it's. Will Gemini do this as well?
- [26:12]
Dan Cumberland
Gemini will do it as well, which makes sense because like in the Google Eco. Yeah, they better do it. If they're doing anything, shouldn't they do that?
- [26:19]
Nathan Berry
But basically it means that these aren't static documents that you create once.
- [26:22]
Dan Cumberland
Exactly.
- [26:23]
Nathan Berry
If you go back and update it or if your other automation puts a new line in the spreadsheet and that then the cloud project knows about it
- [26:32]
Dan Cumberland
and can reference that. Yes, yes. So you're not having to spend as much time going through thinking how like all the documents we just talked about, like, if you have elements of those that are accessible to other team members that are updatable. My product ladder is changing. Like my business is still, is still young. Right. And so I need to go and update that context pretty regularly. But same on the icp. So you could at a company level have these like core context files that are shared amongst many different projects and then someone who owns them, they'll update them periodically and then all the places where they are, where the team is using them gets updated. Which, like, we don't think a lot. I don't hear a lot of people talking about like team design around AI, but I think that's one of the core principles of like, where is your core context for your company? Rather than letting your team just like hack it together and like make their, make their own things. I mean, that's fine and good and there are learnings that come up that way. But there's I think also a, a big missed opportunity of like saying this is our knowledge base. Not even knowledge base, because that has a different meaning in the AI world. This is our core context that should feed your projects.
- [27:32]
Nathan Berry
Yeah, that makes sense. Yeah. Okay, we'll dive into more projects in just a second or like specific examples. But what are some of the big mistakes that you see people make?
- [27:40]
Dan Cumberland
Yes, yes. Okay. So that they. We. We talked about the relying on, on CL on chat. GPT's a memory. Memory, memory function. So that's, so that's, that's one. Another is like thinking that these tools are marketing. Marketing tools. So in my, in my cohort, which is usually it's me, they're doing inside a company with teams in the company. I know you guys have done similar, similar things at Kit. We have someone outside who's in, you know, you guys work with Jay, who's fantastic, come in and help, help the team, you know, learn how to use these tools. I also have a cohort, so I do that, but I have a cohort where, where business leaders join and usually running small teams or they're solopreneurs. But it's A bunch of different businesses, which is kind of like a little bit mayhem because they have different business models in the room. But there's also so much power in seeing how other people are using the tools. And my very first session, we had a CEO of a evaluation team, evaluation company. They do M and A valuations, and I gave a lot of marketing examples throughout that first. We're just talking about, like, you know, basic workflows. The outcome of that, that first session is I want people to walk away saying, here's the thing that I'm going to work on first in my business. And so we got to the end of that and he's like, like, this is all great, but I'm just gonna give it to my marketing team because, like, none of this really impacts me. Um, and I was like, that's exactly what you don't want to do. Uh, it's easy to talk about marketing when we talk about AI because we all know what marketing looks like. We know what the output of marketing looks like, what a post looks like, what a, you know, what a podcast sounds like, et cetera. It's the same for everybody, but when it comes to, like, the internal operations is different for everybody. So it's harder. It's harder to talk about. So one of the mistakes is that people say, well, you know, because that's how we talk about it. That's what, you know, that's all. That is all that is good for. And I think it's a good entry point into the conversation. But I see the real value coming from operations. And there's a handful of teams. So that same, that same individual, by the end of the program, he said, we've increased our capacity so much that we don't have to. We don't have to hire any new team members probably for a couple of years based on the workflows that we built in that program. So going from, like, this isn't relevant to me to, like, I have so much more capacity that, like, maybe now he needs to go back to marketing, right, and throw some. Some rocket fuel on the marketing flywheel, which actually, we're actually working on that with his. With his team. Similarly, I work with Amanda Northcott and her team at Level Upgraders. I know, you know, Amanda, you know, brilliant business designer, and so I've enabled her team, taking her ip, this process that she runs with her clients, helping them build their businesses and helped them AI enable it. So she still has consultants that are working one on one with all the people in her programs, but there's so many pieces along the way where they're having to, you know, they're, they're doing an audit of the whole, whole business where they're, you know, asking all of these questions and then output of that as a report. And there's a lot of, like these, these kind of artifacts that they're building for their clients. So we're able to then take what used to take, you know, days, if not weeks for them, and they can turn it around in a couple hours. And in her words, she's running a consultancy on software margins, which is, you know, unheard of.
- [30:46]
Nathan Berry
Much better.
- [30:47]
Dan Cumberland
So, so, so that is, you know, another, another big mistake is saying that it's just about marketing, but operations, I think, is really where the magic happens.
- [30:56]
Nathan Berry
I love that. So what's the third mistake?
- [30:58]
Dan Cumberland
Yeah, so the third mistake. There's a couple different ways that I could articulate it, but it's really like how you think about one, your workflow and what, what you're doing with AI. And then secondly is about your relationship with AI. And so what you. What the, the remedy to this is bringing attention to both. And what I mean by that, let of them individually first is like, everything that we do has a workflow around it. And so we've all gone into ChatGPT and said, hey, write me a LinkedIn post. And then we've had that. I remember, you know, back in 2022 doing this and like, whoa, that's so cool. And then you do it again. It's like, wait, that sounds just like that last one, right? That's not the. Right, that's not the best process for creating, you know, creating content. So instead you're, you're treating it like, like a done button. Like, you just hit the button and it's done. Instead, treat AI. There's a couple different metaphors I want to use here, but treat it like a sous chef. So you, you need to think about what is, how do you prep the ingredients to create the dish so that your sous chef, you know, you're the, you're the, you're the restaurantpreneur, right? So that your sous chef can come into your restaurant and run your kitchen and cook all the things that you need that it needs to cook. So being really intentional about, about your workflow design and at what points you are interjecting AI into that process. So again, we can talk about creating content. Cause that's the commonality here. Like from IDE to drafting to refining to formatting, to posting, like, there's many Places along that where you could use, have AI. And sometimes you can have it do multiple steps, but you're going to at least want to do like a few of those steps and especially like the last like 5 to 10%. And so this is not just going on, you know, going live and looking like you ready with, with, with AI. So thinking really intentionally about AI, like the workflow, like AI is very much one. It's a soft skill. We think of it like a hard skill, but it's a very soft skill. And it also is a thinking skill. I thought when into this I'd be teaching people like, you know, technical stuff, but it's really like I'm teaching, I'm teaching thinking skills. So that's, so that's one, one, one piece of like how we think about these tools is like a done. A done button instead of a, a, a collaborator or, or a, you know, a teammate. This, the second thing is like, second way I want to think about. I, I like to think about it is, you know, another, the other metaphor is AI is a recent grad. They just, they just graduated Harvard. They took every class and got three days. They got every degree, and today is day one. How do you help them do their job? Which goes back to context engineering. And so thinking really, really specifically about how would I take that collaborator, Like AI isn't your therapist, isn't your buddy. It is a tool, a tool that you need to relate to as if it could do so much, but you have to tell it exactly how to do it.
- [33:51]
Nathan Berry
The analogy that I've heard a lot is exactly what you're saying of if you're delegating to a brand new employee and you're like, do this.
- [33:57]
Dan Cumberland
Yeah.
- [33:57]
Nathan Berry
And you expect it to be done perfectly.
- [33:58]
Dan Cumberland
Yeah.
- [33:59]
Nathan Berry
But like, they have no context, they don't know anything. Now the employee will say, I'd love to do that for you, but I don't know how. The AI will say, no problem. Got it?
- [34:10]
Dan Cumberland
Yes.
- [34:10]
Nathan Berry
And so you have to have that difference of like, okay, if I were to delegate this task to a human, what context and instructions and all that would I give them in order to make sure it happens successfully?
- [34:21]
Dan Cumberland
Yes.
- [34:22]
Nathan Berry
And so it's like I hear you saying is do that with AI. Yeah. Is that right?
- [34:26]
Dan Cumberland
Absolutely. And, and a piece of it too gets really personal. Like how if you were working with another person in that particular part of your business. So let's go back to like, you know, you're a business consultant and you're delivering a report to a client and you have a team member who's very capable. How much of the process do you trust them to do? Like maybe they're a senior team member, they could do the whole thing. But that's because you've done a lot of reps with them. They're brand new team member. You're not delivering, you're not going to deliver the let them own a client deliverable in a client relationship the very first day. Right. And so some of it is like really personal on how much are you comfortable with you know, somebody else owning. And, and I find that there's a very similar parallel like people who are, who have big teams are much more comfortable a lot having AI do more things for them. And so that's one, the one component and the other component is like how, how deep is your relationship around that particular workflow? Like how many times have you done this successfully? And make sure that you have good reps in before you like really lean on it and, and, and send it to the client. Right, right.
- [35:29]
Nathan Berry
The other thing that you said is really splitting it out into separate processes. I see a lot of people say all right, here's my input.
- [35:36]
Dan Cumberland
Yeah.
- [35:37]
Nathan Berry
And the finished result is this.
- [35:38]
Dan Cumberland
Yeah.
- [35:39]
Nathan Berry
And so let's make one project or one automation that does whole thing.
- [35:43]
Dan Cumberland
Yes.
- [35:43]
Nathan Berry
And I, what I hear you doing is chunking it out. In the tick tock example you might have one that is just about pulling the moments and storing all of these moments or, or key stories insights in the podcast. And then you have a library of those.
- [36:00]
Dan Cumberland
Yes.
- [36:00]
Nathan Berry
And then you have the next step that is about taking those and writing them.
- [36:04]
Dan Cumberland
Exactly right hooks. That's exactly right. Yes.
- [36:06]
Nathan Berry
Keep going.
- [36:06]
Dan Cumberland
Yes. Yeah. Another example that I've done, I'm doing for clients right now is around AI search and, and organic search. You know, traditional, traditional search. I hate that we have to talk about that. It takes so many words to say that right now. Like we need just, can we just call it search from now on. It's both things. But this, this really intentional process to take take clients do a technical analysis of their site, recommend improvements that either their content team development or if we, if they have the capabilities we can hook it up to, to AI and have AI implement some of these to get, get, get it positioned for, for search for search rankings and for information in, in search but then also looking at their keywords and then designing this process to draft a post. And it's an eight step process as it stands right now at least for, for one one of my clients that goes from Keyword research, looking at the competitors, looking what they're ranking for their content around each of those. Each of those. Those words. And writing a as much like you would, you would work with a team of writers, a writer's brief of what that article needs to be about, a rough draft, a fact check where it goes and looks at the definitions of all the terms and make sure that it's 100% accurate so there's no hallucinations included. Then there's like a humanization step where that looks at it from like a story arc perspective. And then there's a formatting step. We're looking at the opening and the close in particular. And then there's an AI optimization step where it formats and adds in any extra elements as far as, as, you know, Q and A loves Q and as, you know, FAQs right now and all of those. So that's just another example of, like, how you can take what, what feels like. Yeah, just, just write a blog post for me. AI, like, and, and you can break it into steps. And by breaking it into those steps, you greatly improve the quality of the output because you're giving AI just one thing, one job to do at a time, and it can do that job really, really well, better than probably you could. Right. And then it hands it off to another with a really specific job to create a full workflow.
- [38:16]
Nathan Berry
Well, it's like asking AI, you know, write an article about this with historical examples that is really fun and engaging to read.
- [38:23]
Dan Cumberland
Yes, yes.
- [38:24]
Nathan Berry
And so it's going to be like, okay, and it's going to do that, and then you're going to read it and you're like, I don't think this is true. I said historical examples, and it's going to say, yeah, but you said fun and engaging to read. So I, you know, yes, like, I tried to balance all these. I couldn't do it all.
- [38:37]
Dan Cumberland
Yes.
- [38:38]
Nathan Berry
But then a separate project that says, like, you know, whatever prompt you're giving, like, this is most likely wrong. Find all of the mistakes in this. AI will be like, oh, I can exactly find it. Like, you're making up this.
- [38:52]
Dan Cumberland
This isn't true. Yes.
- [38:53]
Nathan Berry
One that I love is, I'll just say, hey, point out all the logical fallacies that I made in this essay or the, like, leaps that I made without enough data or examples or whatever, and it'll say, you know, okay, well,
- [39:08]
Dan Cumberland
right here, you apart, you know, you
- [39:10]
Nathan Berry
gave one little anecdote that happened in your life and then you extrapolated out to then be this whole thing that now is like a physics level principle that you're telling the reader they should believe and you're like, oh, yeah, I guess I did.
- [39:22]
Dan Cumberland
You know, my conversation with my 10 year old turned into a life principle. Right? Like what? It's just one thing. Thing, right? Yeah. Yes.
- [39:29]
Nathan Berry
And so it's so good at that. And so break like. I think that is a very key principle in AI design is break up the tasks and give them one job to be done.
- [39:37]
Dan Cumberland
Yes. And what I find is the better the output, and then the better the output, the better the impact. And I was telling you about how when I first started experimenting with this AI content, I published 193 articles in one day on my site back in January, and it's 5x the traffic and within the last 12 months we've had had 12 million impressions on, on organic search, which is just nuts. So like, and I was like, I don't know, this might ruin my site. But because I, and that's why I did it on myself, right? I'm like, I'm gonna do it myself and see how it does. And like, okay, this, this works. And so that's why I'm, you know, that's why I'm doing it, you know, for clients now. But you gotta, gotta be the, gotta be the guinea pig to go first. But that's what I find is that the, the more intentionality you bring to the workflow design, the better the output. Whether that output is for the algorithms or for human consumption, that's important.
- [40:26]
Nathan Berry
Okay, so we talked about breaking all these steps up. How do you string them together in an automated way? Because the last thing that we want to do is get to the point where it's like, congratulations, you automated all the fun work.
- [40:37]
Dan Cumberland
Yeah.
- [40:37]
Nathan Berry
And now your job is to take the, the ticket between the different AIs. Oh my gosh, I know so many people that do that. They're like, have an idea and they stick it into one machine and they get it back out and they're like, stick it into this.
- [40:50]
Dan Cumberland
Yes.
- [40:50]
Nathan Berry
And they're like, and now I'm gonna, I like pasted in a kit.
- [40:53]
Dan Cumberland
Yes.
- [40:53]
Nathan Berry
And so how do you avoid being the person from office space who says, I don't know, I, I take the thing from here down to there.
- [41:00]
Dan Cumberland
Yes, yes. Well, I want to hear, I want to hear your thoughts on this because I know there's a lot of different ways that we can do this. And right before we were recording, we were talking, we were talking about how there's different features in ChatGPT on desktop versus versus mobile version.
- [41:11]
Nathan Berry
This is a key tip.
- [41:12]
Dan Cumberland
This is a key tip. Yes, yes on desktop and only on desktop at this moment, which might be deprecated in next week. You can call agents into the chat with the AT sign. So you could have an agent for each step of your workflow and you could work on it all in one chat. You ideate maybe with an ideation interview agent who asks you a bunch of ideas and then you're like, okay, that's a great one. And now at my drafting agent turns that idea into a piece of content. Then at my hook agent, let's make this really like hook worthy at the beginning. At my call to action agent, let's make this like a really solid call to action at the end. And so you can string all these together other in one pain. So that's, that's one way you can also do it. Similarly in Claude with skills and skills is a pretty new, a pretty new feature as of this recording and it is like rocket fuel and nobody knows about it. Uh, people do know about it. But you can teach Claude how to do things. And so what's cool about this is that you can by introducing a skill you can introduce again to. You can just use that content example like teach Claude how to format your hooks and even how, even your process for writing hooks. Like every time I, I'm working on a hook, it give me 10 in and use these templates or et cetera. So it's all built in. Much like you would build a custom GPT or a project or something like that, except it just runs in the background. And then cloud just knows every time you talk about hooks, this is how we're doing it. And so you can build it so that every time, every step of that process, it knows what to do. You could also build in slash commands. And I had this viral post, I think I talked about the viral post earlier about slash commands and chatgpt, which was. It's, it's a, a fun party trick that you can type slash and then a bunch of different words and then ChatGPT will infer your meaning. And it's like a shortcut for like prompting. But in cloud you can actually build them. You can also build those into your custom GPTs. If you're building custom GPTs, you're building a prompt. I highly encourage you to think about how like the different steps that you would want in that prompt and put a slash command next to them so you don't have to like write it out, out. Every single. Every single time. I was talking about my. Just a. This is a super nerdy fitness example. Let's talk about my, my fitness Gemini gem that I, that I use to optimize my CrossFit workout. So I get the little extra that I do after the CrossFit workout, but I upload a photo and then I just say slash add and then it knows. Look at the photo, look at my workout log, look at my training goals, and then tell me what else to add today at the end of my workout. So then I'm not standing here at the gym having like, okay, I just completed this workout and it had blah, blah, blah, blah, blah. And now look at, you know, all it. It's just a much faster way to do that. So you can build those into your prompts in ChatGPT. That's a cool trick. But as far as workflows, again, with Claude, you can build slash commands into Claude so that it always knows what to do no matter what project you're in. So you can just say slash books, slash published a kit, you know, whatever, whatever you wanted to do.
- [44:08]
Nathan Berry
Okay, that's amazing. What about like, I still want to dive deeper into this, tying things together.
- [44:13]
Dan Cumberland
Yeah.
- [44:13]
Nathan Berry
Because that's still relying on me as a person to do it.
- [44:17]
Dan Cumberland
Okay.
- [44:18]
Nathan Berry
And this is something that I haven't solved yet.
- [44:20]
Dan Cumberland
Okay, well, I can solve this for you. Okay.
- [44:22]
Nathan Berry
Let's absolutely do this deep dive. We'll do it for just two minutes and then we'll dive into four more examples that are really practical and like broadly applicable.
- [44:29]
Dan Cumberland
Love it. Love it. Yes. Okay, so, uh, I know that you're familiar with cursor. Uh, cursor is an ide, which is independent development environment. I define that for the, for the, for the listeners that you can know what, what you're looking for when you're searching for things. It's a developer tool. The way that it works, you can think of a window with three panes. Left side is your files, center is your canvas, right side is your AI. You can, you can arrange it any way you want, but that's the typical arrangement. So one way that you can do this is using a tool like this where you have your canvas. These are built for developers to write code, but you can also use them for as non coders. And this is, this is, you know, one of, one of my superpowers, one of my, my, my secret, secret hacks. Using these tools for everything. I run my entire business from cloud code in an ide. And so what that Means is as I'm working on a piece of copy or a report or like whatever it is, I have this canvas and AI can write to that canvas, I can write on that canvas, I can select portions of that and tell AI to change that piece. And so you're not having to do all the copy and pasting until you're ready to copy and paste. There are other ways around that with MCPS and things like that. So that's, that's the technical, you know, CLAUDE code and. Or an IDE is a really great way to do that. You can also connect your cloud in particular, and you can tell me if you know more about the other, the other ecosystems, but to what's called mcps, which is a way to connect data between, send data between AI and apps, but also it can do things in those apps. And this is a little bit more technical, but. But cloud is making this easier and easier. So I could send. Not just pull information, but also push information to the place where I want to push it, which is super powerful.
- [46:12]
Nathan Berry
Yeah. So MCPs are very, very important for anyone who doesn't know an MCP is, stands for Model Context Protocol. And like Kit has an MCP server, a bunch of these. There's more and more coming out all the time. Zapier will go make an MCP server for tools that don't have them.
- [46:27]
Dan Cumberland
Yes.
- [46:28]
Nathan Berry
There's all of this. And so it's very valuable what you're talking about with using Claude code or using cursor in this way. Right. In in particular, I know a few people that do this where I think it's Amir from Humbletics is. Has some great videos on this where he's like, I run the operations of my business out of cursor.
- [46:50]
Dan Cumberland
Yes.
- [46:50]
Nathan Berry
And I have files about accounting and I have files, you know, like taxes, bookkeeping, all of this.
- [46:56]
Dan Cumberland
Exactly. Right.
- [46:57]
Nathan Berry
Processes are all right there. It connects to the QuickBooks MCP server.
- [47:01]
Dan Cumberland
Yes.
- [47:01]
Nathan Berry
And so he's like, I know it's a developer tool, but I'm using Cursor for ops.
- [47:07]
Dan Cumberland
Yes, exactly. And like I'm, I'm using it. I have lead scoring in my Kit account that fires emails into my, into my, my, my Gmail account that says, hey, this person has clicked on a bunch of things. You want to send them an email. And so then I, along with that email, it populates what they have clicked on most recently. And I have an A command in Windsurf, which is like cursors and ide. Or I can do it in Claude that will look at my email, find those and then write them an email based on what they just clicked on from my email account to them to say, hey, I saw. You're looking at, you know, know the cohort. You're looking at your, you know, what, whatever it might be. Do you have any questions?
- [47:48]
Nathan Berry
You could load up 25 drafts in Gmail.
- [47:50]
Dan Cumberland
Yes.
- [47:51]
Nathan Berry
And you could look at them, be like, oh, I actually know this person. Let me add a little detail. Like, hey, congrats on getting married. I also noticed that, you know, whatever thing is. Or you could just have it, you know, yellow it.
- [48:00]
Dan Cumberland
And just sometimes I do that. But. But the more human thing to do is. Yes. To make it. To make it personal. But I give that as an illustration. Like, you can connect these with mtps. You can connect tools to do all kinds of stuff. I love that.
- [48:12]
Nathan Berry
Okay, we promised that we'd get into a few more examples. Okay, what's the next example that you want to share?
- [48:17]
Dan Cumberland
Yeah, yeah. Okay. So just examples of, like, great workflows again. Like, how do we cross the chasm between the AI promise of, you know, not working anymore and, like, where we are right now, where it's like, I've got ChatGPT, but I don't know what to do. There's. There's a handful, a handful that I think are really powerful. We've talked about, like, content, you know, content Flywheel, Another just really, really great one is kind of goes back to, like, thinking about your workflow, but especially around an idea capture system. Everyone should have an idea capture system. I know you've talked about this on the show before about. And it. So it starts with, like, where what. Where is the content being created? And my premise with everybody they work with around this is that we are Scrooge McDucking in content gold, and we're totally blind to it. And that takes me back to my childhood watching DuckTales and Scrooge McDuck swimming in the money. Right? But we're just totally blind to it. We're creating content all the time. But then we go and sit down at our computer and we look at a blink, blinking cursor, we open up LinkedIn. We're like, oh, what am I? What am I gonna post today? Like, you should never have to do that. And so the, The. The invitation here is to think about where are you already creating? What meetings are you having, what calls are you taking, what conversations are you having with your team, with clients, with prospects, and then how can you capture those? Usually with a call recorder? Yep. And really easy workflow around this would be put all your call recorders, your, your, your notes, put it somewhere where it's accessible to your, to your, your AI. You can connect, you know, get them into Google Drive. You can connect your Google Drive with the connector both on chat GPT and on cloud. That's a pretty low, low hanging fruit in, in that regard. But then you can go beyond that of like, how else are you, you know, creating, creating content. How, when, when, when you have ideas, where do you put them? Where do you collect articles? I know you're a big fan of Whisper Flow. I am as well. A week ago I ran an update on my computer and my mic stopped working and I felt like I couldn't use my computer. I was so mad, like, I can't, I can't work.
- [50:12]
Nathan Berry
You're not gonna make me type.
- [50:16]
Dan Cumberland
So, so I got that fixed. It took way too long. But like Whisperflow has like the flow notes, which is just super cool. It's just you hit the record button for people listening. You can have it, this app on your phone where you just record and then it just makes a note, transcribes your voice into a note and it's really easy to then connect that. You know, pull that in whenever you're going to go create content. Grab these notes that you just recorded, copy paste into your, into the place you want to. To put them.
- [50:39]
Nathan Berry
Two things. One, on Whisper Flow on my phone, I added the shortcut to my control panel.
- [50:45]
Dan Cumberland
Oh.
- [50:46]
Nathan Berry
So that the note is created from there.
- [50:47]
Dan Cumberland
Just make it just a tiny bit
- [50:49]
Nathan Berry
easier to create the note. Um, yeah. And. But the other thing is one thing I realized I haven't fixed yet is Whisper Flow doesn't connect to anything else.
- [50:58]
Dan Cumberland
I know. Yes.
- [50:59]
Nathan Berry
So is this.
- [50:59]
Dan Cumberland
Yes.
- [51:00]
Nathan Berry
Is there a way to solve it or.
- [51:01]
Dan Cumberland
I. I have not solved it with Whisper Flow. I use a tool called Audio Pen, which was like a predecessor to Whisper Flow specific specifically for notes. And so it's not as like universal everywhere, but it, it, you can record a note and then tag it and then you can connect those tags as a year.
- [51:16]
Nathan Berry
I'm sure by the time we release this episode, it'll be Whisper Float.
- [51:18]
Dan Cumberland
If you're listening, put this in there. They'll get there soon, I'm sure.
- [51:23]
Nathan Berry
Yep. That makes sense.
- [51:24]
Dan Cumberland
Yeah, yeah, yeah.
- [51:24]
Nathan Berry
So capturing as many capture system.
- [51:26]
Dan Cumberland
Yes.
- [51:26]
Nathan Berry
As possible. So then, okay, if we're using, you know, Fathom a Grain or something like that, by default all those notes are going in a place that you could look at it. But it's not accessible. So you're saying, get that into a Google Drive folder. Where are we putting it?
- [51:42]
Dan Cumberland
Yeah, I would, I'd recommend Google Drive. That's just because it's easy to connect that to all, to all the places if you're, you know, using a different system. Like, you know, we're talking about IDs and, and, and cloud code. I like to actually get them in text files onto my computer because then I can batch process them and do different things like that. But that's like, that's like, you know, the level. Level. A couple levels of that. But yeah, Google Drive is a great place to start.
- [52:04]
Nathan Berry
And so then we would be having a cloud project that is looking in this meetings folder and then analyzing it for, well, let's say in your newsletter you always recommend three products.
- [52:16]
Dan Cumberland
Yeah.
- [52:17]
Nathan Berry
And you're like, I out of products to recommend.
- [52:20]
Dan Cumberland
Totally.
- [52:20]
Nathan Berry
And so it's saying, hey, go through my meeting notes for every meeting that I have. And anytime someone mentions a product, like, add that to the spreadsheet with the reference back to the context.
- [52:30]
Dan Cumberland
Yes, yes. And I'd be a little bit more advanced. Right. Because I don't know that Claude can do that out of the box. Because you can put files into context. I don't know if you can put folders into context, but if you, you know, so that's, that's one thing as far as like a persistent context, but you can call files into the chat. So you could say, you know, you know, at, you know, your last three meetings and say, right, pull, pull content out of the, out of these. And then also the push function would be like needing, needing an MCP to do that. The push function. It's totally possible. You can do that. I do that, but it's a little bit, you know, again, maybe that's level
- [53:05]
Nathan Berry
two or that might be where we need to pull in Zapier in order to.
- [53:08]
Dan Cumberland
Yes.
- [53:10]
Nathan Berry
To run some of this.
- [53:11]
Dan Cumberland
Yes, exactly. Yeah, exactly.
- [53:12]
Nathan Berry
That makes sense. And so then you, you could be mining your meetings for these different content topics.
- [53:18]
Dan Cumberland
Yeah, exactly.
- [53:19]
Nathan Berry
Going from there.
- [53:20]
Dan Cumberland
Yeah, that's right.
- [53:20]
Nathan Berry
It reminds me, Dan Martel, something that he says is he's like, do not create. And I'm like, what?
- [53:27]
Dan Cumberland
That's my whole business. Yeah.
- [53:28]
Nathan Berry
And he says, document, don't create document.
- [53:31]
Dan Cumberland
That's so good.
- [53:32]
Nathan Berry
And so what his whole approach was, he's like, look, I'm running businesses, I'm doing all these things. I'm on meeting meetings. And the approach that he took is he has his webcam in front of him for the meeting. And it's a nice DSLR or whatever, but then he has a separate camera that he set up at a different angle that runs at the exact same time. And he had his editors go through. And so every coaching call that he's on, every single thing that he's doing, he has his editors go through and say, hey, what moments today?
- [53:58]
Dan Cumberland
Oh, that's great.
- [53:59]
Nathan Berry
Like when did I. A client said something, I was like, oh. And I got on high horse and like, yes, like delivered it with energy and passion. I was like, you can do this or whatever. He's like, now I have this in two angles.
- [54:10]
Dan Cumberland
Yeah.
- [54:10]
Nathan Berry
And we're gonna cut out all the other aspects of it and we're going to just use, use that. And so I didn't create content.
- [54:18]
Dan Cumberland
That's wild.
- [54:18]
Nathan Berry
I just documented the thing that I'm doing. Same thing of like having a content team or a videographer come with you on a trip or something. Like I was doing it anyway.
- [54:26]
Dan Cumberland
Yes, we're gonna Gary V. This thing, right?
- [54:28]
Nathan Berry
Yeah, exactly. And so your example is from a writing perspective of like I'm gonna Gary Vee this thing. Yes, but just from my meeting notes.
- [54:36]
Dan Cumberland
Uh huh. Totally, totally. Yeah, I love that. And I think that especially with the video stuff like tools like descriptor, just like their underlord is like insane. Like it's, we're, we're months away from being for that. Like that. Which sounds mind blowing to me to have multiple angles and then having a content team like we're. I'm pretty sure Descript will be able to get us there very soon if it's not already there. I'm not like a Descript power user. Yeah, I love that.
- [54:58]
Nathan Berry
Okay, well, diving in. What is the next example that you want to share?
- [55:01]
Dan Cumberland
Yeah, yeah. Okay.
- [55:02]
Nathan Berry
Okay.
- [55:02]
Dan Cumberland
So just other, just great examples where I've seen just tons of value added. Is using AI as a sounding board especially, especially for, for coaches. Also anyone managing someone to be able to use it as a mirror. And I think AI is a mirror in so many different ways. Like that reflects back to us. Can reflect back to us about really like what it means to be human. Like that's kind of from my, my background as a, you know, started my career in ministry. I have a degree in, in theology and psychology. Like I, I bring this human lens to all of this and so it's really fascinating to see how it's forcing us to draw these lines or at least think about these lines. Like where, where is the line between human and machine. But in. In addition to that, it can also be such a good sounding board or mirror back to us. Our. Our own work. Our. Our work with clients, our work with. With our. Our team members. One friend of ours, who I won't. I won't name him because of the gimp I'm about to share, but. But he's been on this podcast, is a household name and a coach. He was given this example of how he uses it in his coaching practice, and he was working with a client and had a great session all about goal setting and all of this, wrapped it up and felt good about it. And then one of his practices, he takes is. Takes the transcript and runs it through an AI prompt just to help him think about, you know, what. What else did I miss? Like, what, what, what? I think the question that he often asks is, like, what was unsaid? Or what might. Might I have not picked up on? And as you're talking about goal setting, this client mentioned something about. About her brother and. And, like, it was clear there was, like, some pain there. But. But she. But she's like, kind of, you know, brushed it off and then. But he was like, well, we're gonna work on getting these goals done. He's much better than that. Yeah, but I didn't name him, so I guess I can say however I want. But then in the next session, he start. He realized, like, oh, this is like, a major oversight. Like, there's. There probably. There could be a really big unlock here because, like, how much, like, our own limitations, you know, come from our background, our. Our personal lives and these kinds of things. And so he started the session was like, hey, I want to apologize. I think I might have missed something in our last session. As I was reviewing our transcript, I realized that you mentioned your brother and didn't, you know, we. I didn't. I didn't ask more and didn't invite more curiosity around that. And it led to this really productive conversation that took her places that she think really needed to go, but hadn't had someone there to coach her into it. So I get that as just like one maybe more poignant example. Like, this overall concept of, like, AI can really help you not miss the good stuff that sometimes we miss because we're focused on. Focused on other things.
- [57:38]
Nathan Berry
Well, I love that because there's the idea that the more we use AI, the less human will be. And so that's such an example of, like, there was a key moment of human connection and emotional unlock and all these things that came from The AI
- [57:52]
Dan Cumberland
going, excuse me, I think you missed something there. Can I just get on a soapbox about this for a second? Like this AI, because I get, feels like. And again I just mentioned my background. Like I have a degree, a master's degree in theology and psychology and I'm teaching people how to use AI. Like can these things be more different? And it feels like there are opposite sides of the spectrum. But the more we lean into this, I really believe that AI, it can, right? It can, we can use AI to blast junk content into the Internet and spam and spam the whole world. And some people are like, you know, so anti AI because of that. But like who's doing that? It's not the AI, it's the user, right? And so like, yeah, it can drive us to like these non human inhuman practices and things. And like I don't want to like put aside like job displacement and like all those things, all those things are, are real. But for creators and for, for knowledge workers, like even where we started this episode talking about like, about friction, like AI is such an opportunity to get closer to the meaning behind what we're doing. Get there faster with less friction, to lean in deeper because we can offload a lot of the other noise and the cruft that can, can, that can get in the way. And I think of this like in some ways, like it gets me so excited because I think about like in some ways it feels like a return to meaning. Even though it was like what my AI is going to, he's going to make us into robots. A return to meaning that I think, like in some ways I think started like the Industrial Revolution, like putting people into a machine where like you're, you're just a cog and a wheel and an assembly line. And like sometimes it feels like that in my workday where I just have to click on these things and push these things here and do all these things that like a robot could do. So then if a robot can do it, is it really my job? You know, And I know you've thought about this too, like that like, yeah, AI, AI is going to take jobs and it's taking jobs. AI is going to change all of our work. But if we have the, if we're at a privilege enough to have choice over what we're doing and if we're, you know, working in the realms of ideas and knowledge, then I say, you know, give, give ahad the job. Cause that's probably not your job anyways. It probably never was. Is where the Real meaning is. And so while my career feels like it's like these things, I think it's actually a circle. And like these two things that felt so far apart are actually like right next to each other. One last slot on that, just on the Industrial Revolution thing is there's this quote, I think it's from. From Henry Ford about, About, you know, when he's building the. The assembly lines. He's like, the problem with workers is that they come with a brain attached. Like, that's literally the words. And like, to say like, AI is an invitation to like, we don't have to just do the things. Like we don't have a brain and instead can use our brains to do bigger and better things. Yeah.
- [60:48]
Nathan Berry
And it frees up so much to have more output and so much more. I love it. One example on. Because job displacement is. It's going to be a big issue.
- [60:59]
Dan Cumberland
Yeah.
- [60:59]
Nathan Berry
You know, there's more and more companies that we talk to where they're just like, we don't have the same hiring plan that we did before. We don't have, like, we don't need to do all this. And so I'm always thinking, like, is this a good thing?
- [61:11]
Dan Cumberland
Yeah.
- [61:12]
Nathan Berry
And a story that. I don't know if it's real or not, but a story that I heard was, you know, two people watching a construction site and this excavator comes in and digs all of that. And it's one person. And, you know, someone goes, you know, we used to have 10 people do that with shovels and like, those jobs are gone.
- [61:30]
Dan Cumberland
Yeah.
- [61:31]
Nathan Berry
And the first person goes, yeah, it's true. We give those 10, like, we give them spoons instead of shovels. And then it take a hundred people and you realize like, oh, like those
- [61:42]
Dan Cumberland
are just jobs, you know, like, yeah. Yes.
- [61:45]
Nathan Berry
Like, yes, we are now able to create so much more, you know, all this stuff. Because the idea. Yeah. I just always think of what are the things right now that were like digging with a spoon? You know, that equivalent. You're like, you know, you could have a shovel or an excavator. Yes. And like go bigger and bigger. So, yes, the hundred people with spoons are no longer needed. The 10 people shovels are no longer needed. So what are you going to do with that time?
- [62:09]
Dan Cumberland
Yes. And I was thinking about, like, it was very similar, just that idea, like, what. What are you going to do at that time? And it's easy to feel like, but that was my job. Like, I was a designer and I made logos for clients right. And now ChatGPT can do that. Like, there goes my job. Right? But like, it's that, there's that story, which I thought this was where you're gonna go, but of like the, the bricklayers. I don't know who said it. It was like, it's like a famous, you know, parable of like the brick layers were working and someone comes along, three bricklayers, and asks them what they're doing. The first says, I'm, I'm, I'm laying bricks, right? Like, that's the design. Like I, I used to make, I used to design things. And then the next one, you, you, they, they ask him like, what are you, what are you, what are you doing? Like, I'm building a cathedral. And then the last one, he says, what are you doing? Like, I'm building a house for God. And it's like, okay, so the designer. Yeah, you were, you're moving lines in Illustrator, right? Like, okay, you were making a logo. The second one, making a logo for, for a company. Or like, maybe that third one is like, like I'm helping a brand communicate who it is in a visual way to its customers.
- [63:18]
Nathan Berry
And now I might help 10 brands.
- [63:19]
Dan Cumberland
Yeah, yeah. And like, which again, like, I'm not a designer and I know that it can be, you know, these, these, these image generation tools can be really painful and scary in that regard. But like, what if we could zoom out and see like, what is the real work that only I can do? Because there's so much more. I think, I think we're underselling ourselves. If AI is that big of a threat again, and this is talking from really privileged, you know, privileged position, like drive through work. My, my wife, my wife worked the drive through at McDonald's in high school and she loved that job. And like now it's AI and that makes, that makes her sad. And I, I get that. But for many of the rest of us, like, I think we need to, to think bigger.
- [63:56]
Nathan Berry
I think that that moment where we realize reinvention is necessary is so key. So in England in, I guess It'd be the 1800s, there was a, a weaver, right? So he's making like a master of his craft, making all these things. And then in the Industrial revolutions, it starts, you know, these looms come together and it turns into an industrial process. And he had a well paying job that he loses and his family becomes very, very poor.
- [64:25]
Dan Cumberland
Yeah.
- [64:26]
Nathan Berry
And they end up moving to the United States. They immigrate. It's this whole big difficult thing. And so his son talks about this experience and like, what he went through and, and watching his family go from like, probably well off middle class to quite poor and making the sleep and this moment of reinvention. And it's Andrew Carnegie that's telling the story. So the son becomes the richest person in the world.
- [64:51]
Dan Cumberland
Yes. Like in this moment.
- [64:53]
Nathan Berry
And so you think like, the job disappears, you know, and you see like the craft is gone.
- [64:59]
Dan Cumberland
All of this stuff. Yes.
- [65:00]
Nathan Berry
And he takes that firsthand observation and turns it into empire. An empire and all that. And so it's like, what are you going to do with this moment?
- [65:08]
Dan Cumberland
Yes. Oh, man, that's a powerful, powerful story.
- [65:10]
Nathan Berry
Okay, so we promised two more examples.
- [65:12]
Dan Cumberland
Okay, two more examples. Okay. So another example is just thinking about. Well, I, I don't even know how to. What, what, what the, what's the headline here? Let me think about this. Yeah. Another example is like that AI can, can do things that you don't know. Like it gives you powers that you don't even know you have until you try to do this. I have a very practical example from my own, My own life. I was at Craft and Commerce this year. You get. Hayley got on stage and introduced the new intern and then you walked out. Yes. Which was a great moment. Um, and then you talked about how you Vibe coded this app for the, the App Store. And you guys just, just launched the App Store and I, I'd seen the App Store, I was like, oh, this is cool. You can like, you know, find some, some practical tools. I'm not, I like didn't have any that were live yet that I was using. But then you're talking about how you had this idea and you, you, you Vibe coded it, which for people don't know Vibe coding. That means using an IDE or cloud wind surf or, or cl. CL code. I'm losing my words to, to code for you. So you don't write the code but you tell the code the, the, the, the tool what to do. Um, and I was just really inspired by that. I've had the problem with. I do a lot of trainings, webinars. Um, if people join my, join my email list, you'll, you'll, you, you'll, you'll get to some free trainings, um, and like show, show up rates are always a problem. And so like I'm like, you know, using different hack solutions to like try to just. I just want to get this thing onto someone's calendar and it's not as easy as I want it to be. So I was like, if I could just have that button in my, in my emails and there's some hack solutions to do this, it would just be so much easier. So I spent a few hours with, with Windsurf, with using, using AI and build this Add to Calendar app button is literally a button, right? That's live in that app store on, on kit. Now I think, I don't know, you could tell me. I'm maybe one of the first, the, the first Vibe coded user generated apps in the, in the store. And now because I was like, I want to sell this for myself. So I solved it for myself and now anyone can use it. Go, you can go to the app store and you can, it just, you put in your time, start time, end time, you can put a link, put a description and then it generates a Apple Calendar button and a Google Calendar button and an Outlook Calendar button and it's in your email. So people can just click and add it to their email.
- [67:31]
Nathan Berry
So what I love about this is in the old world, yeah, you would have come to me and said, hey, I, I have this obscure little feature request, but in kit, could you build an automated ad account please? For me personally, I promise at least a dozen other people will find this useful.
- [67:48]
Dan Cumberland
Right.
- [67:49]
Nathan Berry
And I would have like prioritized it against like a thousand other things and been like, dan, I'm sorry, but we're never building.
- [67:55]
Dan Cumberland
You don't matter that much.
- [67:57]
Nathan Berry
But in this world, like this is one of our big bets when it comes to AI.
- [68:01]
Dan Cumberland
Yeah.
- [68:01]
Nathan Berry
Is that we want to build a platform where your data is wildly accessible.
- [68:05]
Dan Cumberland
I love that.
- [68:06]
Nathan Berry
So you can pull it into any tool you want, you can push back into it and then you can extend the platform in any way that you want. So when I saw you launch this app, I was overjoyed, like so, so excited because you said, this is a problem for me. Yep, I can solve it. And now someone, someone else who has this problem just says, oh, I just installed the app from the store. It now makes this process really, really easy. You have hundreds and hundreds of installs on this add to Calendar App.
- [68:30]
Dan Cumberland
Thousands, like 6,000 events so far that, that I've counted. I wasn't even tracking for the first like couple months.
- [68:36]
Nathan Berry
So yes. So the thousands of events being added and, and using AI, you're able to extend a product that you use every single day.
- [68:44]
Dan Cumberland
Yes, totally, totally. And I think like the invitation here, like I, I have built things internally for, for customers, built things for myself, but I haven't made something that anyone can Just go and download. And so it was like a little intimidating. And, and so for me it's just an experiment. Like, can I do this? Maybe like the kind of like the content experiment. What happens If I put 200 posts on my, on my, on my blog? But the invitation here is to think about like, what are the things that you. I don't know if I can do this and I don't know if AI can do this, but to spend some dedicated focus time just exploring the possibilities. And I think that's where this could get really practical. Into someone's, you know, someone's specific workflow. Maybe it's a, you know, an automation that you've been thinking about that you really, that you really need. AI is great at building automations. If you want to automate something and make or Zapier N8N or like whatever, start with AI and have AI design it for you and create the specific steps that you need in order to go in and create that automation. So then you go into, you know, whatever the tool is and you have to like, you know, sometimes you have to do like JSON stuff and like, you know, rejects and like all this technical stuff that feels over. Totally overwhelming.
- [69:51]
Nathan Berry
But you ask AI to explain it to you.
- [69:53]
Dan Cumberland
Ask AI to explain to you have a, print a plan and then you just have to follow the plan and soon I'm sure you'll be able to just get operator or, or, or you know, one of these tools that go and like do it for you, but print the plan even then run it through another AI system like we were talking about earlier and have like, is this actually right? Because they're going to find like, oh yeah, that's actually not what that thing is called. It's changed the name and like, whatever. But the invitation is to think about you're. You have so much more capability at your fingertips because of AI. And we're blind to it because we just haven't seen what it can do yet.
- [70:23]
Nathan Berry
Yeah, that's amazing. If anyone wants to go and try to build Kit apps themselves. If you go to Kit University just in your browser or there's a whole tutorial series on how to do it and specifically around Vibe coding apps. Cause it's really, really fun.
- [70:35]
Dan Cumberland
Yes.
- [70:36]
Nathan Berry
All right, so people have probably gotten a ton out of this episode. We've covered a lot of ground. What is one final example?
- [70:42]
Dan Cumberland
Yes. Yes. Okay. Yeah. So many of us have transformations that we're facilitating. Whether we're, we're teaching courses any, any. I mean every business in Some way is taking someone from one state to, to another state. And I think there's a, A, a missed opportunity for a lot of us to document those transformations, especially if we're working with people in coaching settings, courses, communities. And the way that this looks, this way that it can look is to take the, the transcripts or the outputs, the places where people are interacting. For me with my cohort, I do this every, every session I have a doc for each one of my, my cohort participants. And every session I have my AI look at that session and pull out any, any insights or clues, cues about the transformation that's taking place in that session for that person. So then we have where they are at the beginning, the questions that they're asking, the problems we're trying, trying to solve. And then as they progress, we have. They built this workflow. They built this workflow. They saved this much time. They saved this much time. And then by the end, I have this library of like, killer case study content. Anytime I need to write, you know, write a post or something or a case study blog post, whatever newsletter, I can just like, okay, grab a case study. It's all, it's all right there. But then you can also use it especially for like, people who work in less tangible fields. Like when you're, when you're facilitating a transformation, that's, that's harder, harder to quantify. To like have their own words fed back to them about the progress is just so, so powerful. I worked with a coach around this where we, we. She has an intake form where she has people do a self assessment. Then she has a, an outtake, you know, has people go through that same form. And then we have AI compare one to the other and then write a story based on her, her journey. Journey. Yeah, the, the journey along the way so that now when someone's finishing, they're not just like, that was, that was great.
- [72:34]
Nathan Berry
Thanks.
- [72:34]
Dan Cumberland
See you later. It's like, here's where you were, here's where we took you. Here's how these scores on your assessment improved. And then here's how you can keep going on your journey, whether that's with me or elsewhere, which is so, so powerful.
- [72:49]
Nathan Berry
Anyone who is creating content is ultimately in the transformation business.
- [72:53]
Dan Cumberland
Yes, right.
- [72:54]
Nathan Berry
That is what we do.
- [72:55]
Dan Cumberland
Yes. We're.
- [72:56]
Nathan Berry
Whether we're trying to teach you a skill, we're trying to develop you emotionally, or we're trying to take you on a journey, we're trying to create transformation of some kind. And so what you're saying is build the systems to document that.
- [73:05]
Dan Cumberland
Yeah, exactly. So well said.
- [73:07]
Nathan Berry
That's beautiful.
- [73:08]
Dan Cumberland
Yeah.
- [73:08]
Nathan Berry
Dan, this has been fantastic. Where should people go if they want to? They're like, I did not get enough
- [73:14]
Dan Cumberland
hour and 15 minutes.
- [73:15]
Nathan Berry
Like, give me more. I want follow you on LinkedIn. I want to dive in. Where should they go?
- [73:19]
Dan Cumberland
Yeah, well, LinkedIn is great, Dan. Cumberland Labs is my website, but I wanted to, like, have something that's really special for, for your people. So I do these, these roundtable AI discussions. Again, like this idea of like crossing the chasm from where we are to like really that possibility and how it's so helpful to just be able to have conversations with other leaders who are, who are making that crossing that. Crossing that chasm, building those bridges. They're private, they're invite only and they're not recorded. Because I want it to be a safe place where people can be like, we're really struggling and we're not doing anything right or what, whatever. And so we, we do them. We do them about once a month. I bought a domain just for this purpose because I wanted to invite people. It's AI growth roundtable.com okay. By the time this goes live, there'll be a place where you can sign up to one of them. I would love to have, have people jump in. If you want to learn more about context engineering, you can go to dancumberlandlabs.com context those five files we were talking about teaches you how to do all of that. But I would love for folks to join me at a roundtable and dig in together.
- [74:24]
Nathan Berry
AI growth roundtable.com that's right. Perfect. Thanks so much for coming on.
- [74:28]
Dan Cumberland
Thank you for having me. This has been so fun.
- [74:30]
Nathan Berry
If you enjoyed this episode, go to YouTube and search the Nathan Berry Show. Then hit subscribe and make sure to like the video and drop a comment. I'd love to hear what some of your favorite parts of the video were. And also just who else do you think we should have on the show? Thank you so much for listening.