The State of AI in Marketing (Report) & What's Coming Next

I’ve watched AI move from an interesting experiment to a daily marketing tool faster than most people expected. Two years ago, marketers were still asking whether ChatGPT could write useful copy. Now the more relevant question is: what happens when AI touches nearly every part of the workflow?

That’s why I spent time with Social Media Examiner’s State of AI in Marketing report. The report offers a useful snapshot of where marketers are today, but the more valuable clues point toward what comes next.

My biggest takeaway is simple: AI adoption is no longer the main advantage. Most marketers are already using it. The advantage will come from knowing what to automate, what to leave human, and which bottlenecks are actually limiting growth.

AI use has crossed into the mainstream

The report found that 73% of marketers use AI daily. Another 17% use it at least weekly.

That’s a major shift. We’ve moved past the early-adopter phase and into the early majority. I predicted that marketing would cross this adoption chasm in 2025, and the numbers suggest that it happened.

If you use AI every day, you’re not an outlier anymore. You’re in the majority. If you only use it once or twice a week, you’re increasingly behind the curve—not because you need to chase every new tool, but because regular AI use is becoming part of how marketing work gets done.

The momentum is still building:

• 84% of marketers increased their AI use over the past year.
• 78% expect to use AI even more over the next 12 months.
• About one in three marketers say AI now touches most of their work.

That last point matters most. AI isn’t confined to brainstorming anymore. It’s showing up in strategy, copywriting, social media, research, internal operations, campaign execution, and analysis.

I use AI in a similar way. It’s less like a search engine and more like a consultant I can call whenever I need help thinking through a problem. I’ve used it for everything from marketing decisions to troubleshooting a garage door opener. The difference is that modern AI doesn’t just find information. Increasingly, it can help do the work.

Marketers are becoming more sophisticated AI users

The report also suggests that two years of AI experience is now the most common level for marketers.

That changes the nature of the conversation. Early on, people were mostly asking whether AI could produce a decent answer. Then they started learning how to provide better context and steer the output. Now many marketers are moving beyond the chatbot interface entirely.

In my circles, conversations that once sounded advanced are becoming normal. People talk about building personal operating systems, installing agent tools, connecting AI to their files, and using coding environments to complete real projects.

That doesn’t mean everyone has mastered these systems. The technology changes too quickly for that. Even people who have been experimenting with AI for years can feel like beginners after a major shift.

That’s actually good news. The field keeps resetting. Skills from prompt engineering still help, but prompt engineering itself is no longer the center of the game. We’re moving toward skills, loops, tools, and orchestration—systems that allow multiple AI actions to work together on a larger project.

I’ve written before about the step that creates more consistency with AI. The same principle applies here: better outputs usually come from better systems and better context, not from searching for a magic prompt.

Claude, ChatGPT, and the platform race

One of the report’s more interesting findings concerns which AI platform marketers value most.

When asked which tool they would keep if they could only choose one, 42% selected Claude and 39% selected ChatGPT. That’s a meaningful change from the previous year, when ChatGPT had a more obvious lead.

I use Claude constantly, and it does a lot of work for me. Still, I’m stubbornly a ChatGPT and Codex user. If I could only keep one platform, I’d probably still choose ChatGPT because of how well it fits the way I work.

But the specific winner matters less than the broader trend. The tools are becoming more comparable, while their surrounding ecosystems are becoming more important.

Google is a good example. It has enormous resources, infrastructure, and an ecosystem that includes Gmail, Docs, Sheets, Drive, and Workspace. Yet it still hasn’t integrated agents into that ecosystem in a truly meaningful way. If Google eventually turns Workspace into a coordinated agent platform, it could become a serious force very quickly.

I wouldn’t count Google out. But for now, the most useful question isn’t “Which model is best?” It’s “Which environment helps me complete valuable work reliably?”

The next learning curve is agents

Most marketers are no longer trying to understand what a reasoning model is. Reasoning has become table stakes. The new conversation is about agents.

Agents are systems that can take actions rather than simply respond to a prompt. They can use tools, work through files, interact with software, execute multiple steps, and check their own progress.

Agents have been overhyped for years, but they’ve only recently started becoming consistently useful. Tools such as Codex, Claude Code, Cursor, OpenClaw, and Hermes are early examples of this shift.

The important breakthrough wasn’t only better reasoning. It was the development of better harnesses.

Why the harness matters

A harness is the environment wrapped around an AI model. It gives the model access to the right tools, files, permissions, memory, checks, and interaction patterns.

A chatbot can give you instructions for changing a file. A harness can potentially open the file, make the change, test it, and report back. That’s a completely different relationship with AI.

Even standard chatbot applications are gradually adding harness-like features. They can create documents, organize information, connect to services, and work with more of the user’s context than they could before.

Over time, we probably won’t talk about harnesses any more than we talk about reasoning models today. They’ll simply be part of the expected AI experience.

If you haven’t experimented with AI outside the normal chatbot interface, now is a good time to start. My look at AI agents and custom GPTs is a useful starting point for thinking about how these systems change marketing workflows.

Automation creates a new problem: bottlenecks

Here’s where I think the next major challenge will appear.

As marketers become better at building AI systems, they’ll automate individual tasks faster than the rest of the process can handle them. That creates bottlenecks.

Manufacturing faced the same problem when companies introduced robots. A robot could make one stage of production extremely efficient, but the next stage might not be able to keep up. Half-finished inventory would pile up between the two stages.

You can create the same problem in marketing. An AI system might generate hundreds of social posts, but nobody has time to review them. It might produce campaign concepts faster than the team can launch them. It might automate reporting while the real issue is that nobody knows which metric should drive a decision.

Automation can make you feel productive while moving the constraint somewhere else.

That’s why the next competitive skill won’t just be building systems. It will be finding the part of the system that is actually limiting growth.

There are always more things a marketing team could do: launch another channel, improve SEO, create more content, run more paid campaigns, segment the email list, produce more video, or redesign the funnel. But usually only one or two constraints are truly holding the business back.

More activity won’t fix the wrong constraint.

Use the algorithm before you automate

The framework I keep coming back to is the five-step algorithm used in Elon Musk’s companies. It’s a practical way to improve a process before handing it to software or AI.

1. Question every requirement. Don’t assume a requirement is valid simply because someone has repeated it for years. Ask what must be true and why.
2. Delete every possible step. Remove unnecessary work before trying to make it faster. If you automate a bad step, you’ve only made the bad step more efficient.
3. Simplify and optimize. Once the unnecessary parts are gone, make the remaining process easier and clearer.
4. Accelerate cycle time. Reduce the time it takes to move from one stage to the next. Speed matters after the process makes sense.
5. Automate or build software. Only now should you give the process to AI, an agent, or a software application.

The order matters. AI makes automation so easy that it tempts us to start with step five. That’s often backwards.

Do the work manually first

I recently talked with someone considering building an app for a new service. The app might eventually be a good idea, but I suggested starting manually.

Text the clients. Have them send information by email. Use a Google Sheet. Work through the awkward parts yourself. Find out where people get confused and where the process breaks.

Only after that should you build the software.

The McDonald brothers used a similar approach when developing their fast-food system. They physically walked through the kitchen process, using a mock setup to test movement and timing before they tried to scale it.

Software is just another form of automation. If you don’t understand the manual process, you won’t know what the software should do. You’ll simply encode your assumptions—and AI will happily build around them.

Where AI marketing is heading next

A framework from OpenAI describes five stages of AI development:

1. Chatbots: conversational systems that respond to questions.
2. Reasoners: systems capable of human-level problem solving.
3. Agents: systems that can take actions.
4. Innovators: systems that can help invent and discover.
5. Organizations: systems capable of doing the work of an entire organization.

We’ve clearly passed the chatbot stage and moved through the reasoning stage. We’re now in the middle of the agent stage. My prediction is that agents will become much more mature over the next year.

After that, the most interesting shift will be AI that can question assumptions and contribute to invention. Current AI can reason through a problem, but it often accepts the requirements it’s given. It doesn’t naturally step back and ask whether the entire premise is wrong.

That ability to challenge assumptions is essential for innovation. It’s also one of the reasons human judgment still matters.

AI can help with pattern recognition, execution, research, and iteration. Humans still bring context, taste, empathy, responsibility, trust, and the ability to understand what a situation means beyond the available data. Those qualities won’t disappear just because agents become more capable.

You can read more about the human edge marketers need to preserve as AI takes on more execution.

What marketers should do now

Most marketers are focused on learning how to build AI systems. You should learn that too. But if you want to get ahead, add another skill: bottleneck identification.

• Map the entire workflow, not just the task you want to automate.
• Find where work piles up or decisions slow down.
• Question the assumptions behind each requirement.
• Run the process manually before turning it into software.
• Automate only after simplifying and testing the process.
• Keep human judgment visible where trust, taste, and responsibility matter.

AI adoption is spreading quickly, but more AI doesn’t automatically create more growth. The marketers who stand out will be the ones who can see the whole system, identify the real constraint, and then use AI to improve the right part of it.

Start by finding the bottleneck in your current marketing process. Then apply the five steps in order. That’s a much better place to begin than adding another shiny tool.

The State of AI in Marketing report by Social Media Examiner reveals the widespread adoption of AI in marketing, with 73% of marketers using AI daily. The report also highlights the emergence of AI agents and the need to address bottlenecks in AI-driven processes. Looking ahead, the report suggests a future where AI innovates and even runs entire organizations. AI Society (by Social Media Examiner) State of AI Report: https://aisociety.socialmediaexaminer.com/2026-ai-report/ Takeaways Widespread Adoption: 73% of marketers use AI daily Emergence of AI Agents: Addressing bottlenecks in AI-driven processes Future of AI: AI innovation and potential to run entire organizations Chapters 00:00 The State of AI in Marketing 02:32 Adoption and Daily Use of AI 05:22 The Maturing Experience with AI 16:05 Addressing Bottlenecks in AI Processes 27:59 The Future of AI in Marketing

Dan Sanchez: Hello and welcome back to the AI Driven Marketer. I'm Dan Sanchez. My friends call me Danchez. And today we're gonna give a state of AI in marketing report. Most of the because my friends over at Social Media Examiner have dropped their industry report, which is really good. But I'm gonna be giving my two cents as we kind of casually go through the report. Of course, I'll leave a link in the show note for you to view it. But more important than talking about where AI is today and maybe where we've come from if you're new to this AI and marketing. pursuit is talking about where it's going. Cause that's where it's really at, right? It's not just about where it's where it is and what we're all trying to learn now, but how can we get ahead of the curve. In marketing, there's always an advantage to thinking two steps ahead of everybody else so that you can actually take advantage of things that are to come rather than just scrambling to catch up. So stay tuned as we go through what's going on today, where we've come from And I'm gonna be giving my two cents about where we're going. And I'm not just making it up either. It's not just Dan forecasts. I'm pretty confident this is where we're going based on some historical trends we've seen before in AI and in other industries. So let's dive into it. Starting with social media examiners or AI societies 2026 AI Marketing Industry Report. And let me go ahead and share my screen. You could see they actually ungated this report, I think, for the first time. So if you want to find this report, you can just go to social mediaexaminer.com. You'll find a link and jump straight to it. I can tell this one was made by Claude Design, I'm pretty confident, but it looks freaking amazing. Like it's much better than their other reports because AI's just gotten better at making reports look good. and I know Mike is a huge fan of Claude and Claude Design and it it it shows up. So we'll skip past the intro and kind of just dive into the meat. I don't know about you, but I like visuals in these reports. And here's one of the big ones is that seventy-three percent of marketers now use AI daily. Man, if you've been listening to the show for a while, you remember me predicting even two years ago that we would cross the chasm in 2025. And we did, and it happened. We definitely crossed the chasm of most marketers using it on a daily basis. early adopters to early majority. That's the chrasm we dropped last we that's the chasm we jumped over last year. But now it is picking up steam. We're gonna roll over the adoption curve and look at that. We're already at seventy three percent daily use. And seventeen percent are at least at weekly use. And of course there's always laggards. There's always a few people. You know there's that two percent that never use it and the two percent are probably the people that are like offended by it. The people that are really romantic against it. You know, probab and that's it's fine. It's just and I get their reasoning for it, but at the same time I'm like, man, sometimes you just gotta get on the train. But it's you know, they're free to they're free to protest. That's fine. I don't like what some of the things are coming down the pike, when it comes to AI either. But it is a tool. It certainly makes work easier in some ways, harder in other ways. But I think it is important to know that if you're using AI daily, you're in the majority. That's that's most people. So we gotta be thinking about again what we gotta do to get ahead. If we're using it only weekly, a couple of times a week, well, you're you're behind. This this seventeen percent of you behind. If you're listening to the show, you're probably well on the seventy three percent though. So you're probably good. Let's move on. Eighty four percent increased their AI over the last past year. So again, that's why we've jumped from like just about fifty percent last year to eighty like what was it? Seventy three percent this year. It's it's growing all the time and seventy eight percent expect to use AI even more over the next twelve months. So this is gonna grow. This is gonna continue. Pretty soon, daily use. We're gonna start measuring in hourly use. And if you're like me, you're using it multiple times a day, all day for all kinds of areas of your life. It's like it's it's such a helpful thing to have like a pro consultant in your pocket whenever you need to think about anything. Health, legal, fixing your garage door opener. Believe me, I've gone rounds on that one. all kinds of things. It's just so helpful. Like we're moving on past Google for sure, because we're finding that AI, especially now with its just agentic ability, is not able just able to find the right answer and coach us through something, but also able to help us do the work. And that's the big deal now. Bulls, let's move on. Marketers are experiencing two years in the is the most common level of AI experience. So last year I remember this report said like it was it was everybody had about one year of experience. Everybody was still relatively new, but you know, it's maturing now. Like people are starting to have more than a year. We've we become comfortable with the AI behaves and becoming more nuanced in how to lead it in certain directions. That's becoming more the norm now. Before it was kinda like, I don't know, it just doesn't work sometimes and people would get confused when it would make stuff up sometimes. When they didn't give it enough context, for example. But now it's it's becoming much more normal for marketers to have two, three, four years of of experience with AI. So we're becoming much more sophisticated as a whole. I know every time I'm hanging out with just like random people, oftentimes AI is a point of conversation and people aren't just talking about like, yeah, they use Chat GPT to answer a question. Oftentimes the question is like, hey, I installed Hermes, I installed OpenClaw, I'm creating my own OS. These are the normal conversations I'm having with just normal people I'm bumping into now. Before it was like, Hell have you tried that? yeah, I just installed it and tried it. That was the conversation two years ago. A year ago, people were just talking about like, yeah, I'm using it more and more. And now it's like, I'm it making it my OS. Yeah, that's that's becoming normal conversation, at least in my circles. And I'm not even talking about talking about people who are like other AI enthusiasts like I am. And then it usually comes up in the conversation like, Hey Dan, have you tried AI? And I always smile because they don't know that I have a whole podcast on this topic. It's fun. It's fun. one of the things it says for one in three marketers, AI now touches the majority of the work, right? If like using it for everything. You're using it to help you do copy, do social, strategy, tactics, execution, doing internal stuff all the time, right? It's like it's touching it almost a little bit of everything, if not fully taking over some aspects, right? And this was one of the more let's see, there was one stat in here that was kind of interesting. It says Claude edges out chat GPT is the most important platform. This is probably the biggest mover when it comes to the tech what for marketers is that Claude is now like if you could only if we essentially they surveyed all these marketers and they said like if you could only keep one, what would it be? And 42% said Claude and only 39% said chat GPT. And that was a Big shift from last year. it's wild that it's it's gone that far with Claude. I'm actually I use Claude all the time, but I'm still, still, stubbornly a chat GPT guy. I I still think chat GPT and Codex is generally a better tool for marketers. if you like but it's honestly it's kinda like whatever. Like I'm using Claude all the time too, and Fable's amazing. if I can only pick one, it's still gonna be chat for me, and that's fine. It's fine. I'm I'm really enjoying Claude. It gets a lot of work done for me. A lot of their features from one to one are pretty comparable now. less the smaller players are still smallest players. It's interesting. We thought Google would really take a big chunk of the market. You know, they were coming in hot last year at 2025. And it was kind of like she was chat, GPT and and Gemini from Google. And we're like, ooh, who's gonna take off next year? And Google essentially fell behind and they still haven't really shipped a meaningful update since December of last year, which is just crazy. It's like years behind in AI time. So We'll see though. Google's no slouch. They are still they've become a sleeping giant again. They were kind of in the race and now they've become a sleeping giant again. But don't discount them because it's Google. Lots of resources and infrastructure. And of course they have the whole ecosystem. What's crazy to me is they haven't really taken advantage of the whole ecosystem yet, right? They haven't actually like started integrating all their tools with agents in a really meaningful way yet. But once they do and Google workspace becomes this like master platform. You know, it has some real potential. So I'm still looking forward to that. and that's kind of where that's mostly what I want to review in this report. Man, there's a lot of stuff about how people are paying for tools, content marketers create with AI. It's it's just there's so much in this report, but I only wanted to review those few stats to kind of give you an overview of where we are today. If you want to get the report, go to social mediaexaminer dot com or I will link to this report in the show notes for you. To get a full read on, or just to grab the link and, you know, give it to your AI. And I highly recommend that. Even if you glance through it yourself, like give give the link to your AI and be like, hey, based on things we've been talking about recently, what are some stats in this report that are helpful for me? Highly recommend because your AI knows what you've been talking about. It knows you pretty well. Like ask it to highlight the things that you sh you should know and it take take advantage of. So quick tip there. But if there's one big takeaway from this report for me, it's that what they want to learn. And it kind of gives you an idea of where things are going over the next six months. Okay. So that's what we're gonna dive a little bit deeper into, is where where we're at right now and kind of where it's going. And I'm gonna look I'm gonna put my my long side out and look a little farther ahead here and tell you what's coming after that. And I think that's the important thing we need to keep in check as a marketer's if we want to get ahead. We can't just learn what people are trying to learn now. We have to think one or two steps ahead. So the thing that this report talked a lot about is that people are mostly wanting to learn, not about AI. They're all in AI. We're all talking to a chat bot and like we're way beyond trying to learn about how to optimize reasoning models. We don't even talk about reasoning models anymore. It was all the hype a year or two ago, and now it's just kind of dead because it's so good at reasoning, we just forgot that that was used to be a feature and now it's just table stakes. no one's thinking about that anymore. Now we're talking about agents, right? Because for years we hyped up agents and, you know, they've really only just arrived over the last six months. I'd say open claw was kind of the beginning of that. Of course we have you always have like pre signs of it working and Cursor with code and all that kind of stuff. And Claude Code was starting to come in. But it was really wasn't until January when Claude Code really started taking on prominence for more than just code and open claw started becoming a thing that agents really started taking off and becoming helpful and actually reliable. I'd say somewhere around it March, April, especially May, these agents actually started getting reliable to the point because we figured out something over just the last six months, and that is the power of the Harness, which is essentially the the the wrapper or container we put AI in. It's not enough to have strong AI is what we figured out. We can make AI really intelligent, but we also have to equip it with the right tools and the right kind of platform to be able to do what it needs to do. And now that it's become a good at reasoning and getting things right and finding the answer and taking action reliably, it needed to be given a proper container in order to do it. So it wouldn't mess things up, but actually could take advantage of the context and proactively looking for the right answers and taking action in the right things in the right places at the right times and checking in in with you in the right amount. That's essentially the the harness or the container we put AI in in order to get things done. And we found that the power of the harness is actu was like the the unlock that we needed for agents to actually be useful. Of course, reasoning was a really important part, but that harness thing was really the last step. And so what's a harness again? A harness is essentially clawed code or codex or cursor as a harness, right? It's like these other software programs wrapped around it. And if you haven't gone deep into that, I highly recommend checking out my podcast episode from a few months ago about switching from chat GPT to codex. That's essentially taking it from chatbot to using it in a harness. And that's the big difference with using codex as a Mac app on your computer. Is you're using it more like that. Now more and more the the just like we don't talk about reasoning anymore, well, we won't be talking about harnesses anymore. That'll just be like the norm. It'll be just table stakes. In fact, ChatGPT made that big update a month ago and it made even just the app and the iPhone app and the web app to be a little bit more like a harness. Not quite as full featured as something like Codex on your computer where it can take advantage of the tools on your Mac and work through the terminal and do all kinds of fun stuff. But it still had more features. You could save documents, it can make documents, it could rearrange things and do and and use connectors to check your Gmail and do quite a bit more than it used to before. So even your normal chatbot apps are becoming more of a a proper harness. so that that that will continue to blend over the next couple of years. But still, we know these things are out there. And where most marketers are at is we've gotten good at using AI as a chat bot, but not everybody's confident in how to use it as an agent yet. How to take advantage of these these harness features with AI automating a lot of your work. Where's a lot of people are dabbling with that. And we all are, because again, like unless you were on the cutting edge of twenty twenty five last year, like this stuff has all been new to you and it's only been out for a few months. Right. If you've been in OpenClaw, you would have been like one of the OGs who tr tested it first in January when it was still unreliable. but now more and more people are trying it. More people are trying Hermes, the competitor, right? more people are using Codex now. I've been in Codex since April and have been swinging at it, but we're all pretty new to the game, which is fun. If you're here and you feel like you just got into it this year or you've only been into it two years but you haven't taken a heavy swing at it, just know that even the people who have been in it, like me, it it it changed so fast over the last six months that I I feel like I'm not a beginner But I feel like it levels the playing field. And I'm pretty sure that will continue to happen a few more times. It's not a slow thing. Like within a few months, things can change rapidly when it comes to AI. So now would be a good time to dig in because it does compound a little bit. The things I learned about prompt engineering does help me now, but not in the way that it used to. Prompt engineering is essentially gone away and we don't even think about that anymore. Now it's about skills and loops. And about orchestrating multiple agents to complete larger projects for you. and that's where it's really starting to get exciting. But that's the thing people are trying to learn now. And it's not something I'm gonna dive into on this particular episode, but that's where people are at now. They're trying to figure out how to scale systems, not just get one task done, not just get a few tasks done, but how to automate real work. And I think this is interesting because this is where we're going to run into the problem. And I've talked about this in past episodes, but This is where people are at now. So where are they going to be in the future? Let me tell you, because it's already happening now and it's going to happen way more in the future. As people become more proficient at setting up systems and automating work, we're going to run into a whole new problem. And then people will try to be figuring out this problem. So if you want to be ahead, listen to me now. We've run into this before. We ran into this with manufacturing in the 80s when we started bringing in robots to automate human labor tasks in manufacturing plants. And we all hoped, at least back then, I can't say we 'cause I was only born in the eighties, but they hoped that bringing in these robots would make the work more efficient. And it did, and it did speed up production, but it created problems because you can only speed up so many parts of the this whole system. And no one, even till this day, has not created a fully robot driven manufacturing plant. It just doesn't exist. and every time they've tried, even Elon has taken a swing at creating a fully robot driven manufacturing plant and found out that it breaks, it doesn't work, and you get behind. So the problem that occurred is they ran into these things called constraints, where a robot could get things done and then it would pile up on the other side of the robot because you became really efficient in one place and unefficient in the next place. So it would create something we call bottlenecks. Inventory would pile up on the next part after it maybe got finished in stage one, it would get queued up for stage two. And stage one was running so fast all the time, you just have a pile of half finished inventory at stage two because stage two could only go so fast. You couldn't automate it. It couldn't be as efficient. That's the same thing we're running into now. People are using AI and getting really cool with Hermes and OpenClaw and Codex and Cloud Code and doing some really cool stuff and automating things that shouldn't be automated or just busy doing work that doesn't really move the needle. And we're finding that more and more. And people are starting to report on it. I called it before I actually started seeing other people talk about this. but we're going to run into this bottleneck issue. So what's the next thing people are going figure out? They're going to figure out how to find bottlenecks and actually focus all their effort and all their AI on fixing the bottleneck. More and more this is going to be part of the conversation. The cool thing is is that This won't stop being part of the conversation. Because even when they figured this out for manufacturing, and there was one book, I can't remember the name of the author, but he wrote a very popular book, probably a million copies sold for a for a nonfiction book. That's a lot of copies. of a book called The Goal, which talked about the con theory of constraints and bottlenecks and how it happened in manufacturing. It got read by a ton of people and they were like, this is the best thing ever. I love this book. It's almost like you were reading my mail, like you've been sitting on my floor with me. And your solutions, they make so much sense. It's almost common sense. So like you knew he hit a nerve. And then he would the author would ask the people who had read the book, like, cool, like what have you done about it? And they'd be like, Nothing. He's like, Maybe two percent of the people who have read the book have actually done something about it. That tells me that we're likely, because human behavior hasn't changed too much in 30 years, human behavior is probably going to hit the same. Which means we're probably going to need a lot of human expertise and consultants to help people to find out what problem they actually need to be addressing in order to grow. Because we all know growth, there's a lot of things you could do to grow a company. You a lot of things you could be doing more in marketing. You could be doing more social. You could be activating new channels, new podcasts. New SEO campaigns, new SEO audits, new we could do paid we could do postcards, we could we can go segment our list better. There's like all these freaking things. Everyone's got ideas. Everyone. Your mom, your CEO, the the the most junior intern on your team, and your senior advisors. Like they all have ideas for marketing. Not all of them are created equal. The trick is you need to figure out what's actually stopping your growth. There's probably one or two things that are actually holding you back. So no matter how much you optimize everything else, these two things are gonna be limit limitations as far as how you grow. And I don't care whether you're in B to B enterprise sales or you're selling, you know, a thirty dollar product on e commerce. It's the same thing. It's the same problem. There's a few key problems that I can't tell you what it is, but there's a few areas you can be looking, and it might be you. It might be you, which how is AI gonna fix you if you're the bottleneck that keeps work from getting done in your company? Could be your boss. It could be a technology technology problem. It could be a funnel problem. It could be a paid media problem. Maybe you're just not spending enough. Maybe you're not doing enough creative. I don't know what it is for you, but it's there and it's somewhere. The problem is people are getting really efficient with all these AIs, just like robots on the manufacturing floor and They're going to be doing this more and more. Again, right in the early stages of using agents to make work more efficient. And people feel super productive doing it. I promise. More and more you're going to start to see this idea on the theory of constraints and bottlenecks pop up. The goal is going to go up. The new book that just hit the market that was fantastic and kind of speaks to the same thing, but is is the more I guess it's the more practical version of the goal and it's much shorter because the goal is a long book. Fantastic book though. Highly recommend reading. It's all s it's all one story, so it's really enjoyable. The audiobook, they like put is there's like music in it, voice actors, and again it's a story. The goal. It's a crappy cover. Listen to it though. It's a fantastic audiobook. Great for road trips. listen to it. But the more the the up to date version of the goal, the more practical. This is what you need to do. Here's the steps is called the algorithm. And it's was written just recently and published by the guy. Who was the president of Tesla? So this is essentially Elon's algorithm for making companies not only efficient but effective. And I think the algorithm, those steps are going to become more and more practical for those using AI. Let me read them to you real quick because I think it's worth just a glance real quick. But I think this is really going to become a very important playbook for knowledge workers and especially marketers who are using AI to do do the work of marketing for a company. But here's the algorithm from Elon's companies that now it's funny, because it's become I've even seen employees from SpaceX and all these places like leave and essentially use the algorithm to create disruptive companies. And it's awesome to see. whether you love or hate the algorithm is quite effective. Here's the algorithm. Five steps. One, question every requirement. You need to get rid of things in order to focus on the things that actually matter. That's one of the things that could be holding you back. also you have to question question the requirements, you might be thinking, I we have to do it this way, but maybe not. Maybe you can do it a totally different way. Maybe you thought, you know, you could only post so much. Maybe it's way more, maybe it's way less. You don't really know, but you need to question the requirements and actually look for the facts. We all have underlying beliefs and we all have underlying beliefs we don't even think to question. This is one of the biggest dangers when it comes to AI because AI will only run with the beliefs you already have. AI will often take your requirements and then build the wonkiest crap. Because you over you automated the system because you gave it the requirements, which is a good thing to do when you're handing off a project to a person or a or a machine to execute. You want to give it parameters, right? But what if the one of those parameters are a fatal assumption? Well, AI is going to bend over backwards to make sure to accommodate for it, and it's not quite smart enough yet to question your requirements. It'll be kind of scary when it is smart enough to question your requirements, and I don't think we'll like that very much, but honestly, it's one of the things that is still making humans human. Because a human can question it. An AI just takes the orders and runs with it. Unless you ask it to question it. But maybe we should start with that. So number one is question every requirement. Number two is delete every possible step. The only way you can do that too, generally, is by doing it yourself manually first. And they cover this in the book and it was really good point. You don't wanna automate first, you wanna automate last, which is why We even though we're on step two, automates actually step five, it's the last option. Yet with AI, it ends up becoming the first option because it's easy to do with AI. But it's hard to question the requirements. It's hard to delete every possible step until you've worked through it manually. Even the McDonald brothers, if you've seen the movie The Founder, the the McDonald brothers, when they were inventing the McDonald system and how to get burgers out fast, like almost like role-played cooking in the kitchen with fake s with fake spatulas and fake burners trying to optimize movement from one place to another in order to make the most efficient way to get hamburgers across, but they had to do it manually first. It couldn't just be an idea that someone drafted on a napkin or even on some really nice graph paper back then. I don't know what they did back then for drafting processes like they had to manually go through it and do it manually over and over again before they could do it. step three is to simplify and optimize. That's part of that manual process of figuring out where things could be easier and then improved little by little by little. Number four is accelerate cycle time. Speed is the name of the game. And if you can take a complex process and simplify and speed it up, that makes a big difference for a company. If you have a regular promotion cycle you're doing, and you can speed it up, improve it and optimize it and make it really fast to deliver it, while now you're freed up to do extra l better things and maybe get better results of out of it. And then last, number five, is automate. Those are the five steps to the algorithm. And it's kind of the new way to deal with constraints and deal with problems you're running into. And I think is like the best list that I've seen with kinda when it comes to dealing with bottlenecks. But again, bottlenecks have been known and people don't do anything about it. So this is gonna be a big deal for a long time. As people start to get into systems and how to do things with AI, you need to learn that too. That's important. That's what people are learning now. So scramble to learn that. And then scramble to master the algorithm. Learn how to do things the hard way, do things manually, and then roll it out to AI. I was talking to a friend who's thinking about building an app. And I'm like, mm. I think an app is a really good idea for what you want to do, but you should probably do it with your clients manually first and work through all the bugs to simplify the process and then you'll know what needs to go in the app. But if you start with the app, the app's probably not gonna be a great experience. If you want it to be a killer app, start by doing the process manually. Text people, have them send you things manually. Come up with a Google sheet for them to enter it into. Do not automate the process. By the way so creating software is just another way to automate a process. So consider that you could put software on the last part of that line too. Number five is automate slash make software. That's the problem. Now I know what you might be thinking, Dan, what about what's after that? What's after bottlenecks? What do we hit? Let me bring you back to this graphic OpenAI posted. Actually, maybe this came from Bloomberg, but this is reported on from OpenAI. And it kind of shows the progressions of the different stages of artificial intelligence. Chet eight OpenAI put this out two years ago, and it's kind of held up. It's been true so far. And I think it's honestly where we're going. So let's talk about where we're at here and then talk about what's next after we run into after we run into these bottlenecks created by agents. And we we've already hit level one, chat bots, AI, conversational language. We hit that as soon as ChatGPT hit the market. Level two is reasoners, human level problem solving. We started seeing that come out two years ago almost. And it got perfected last year to the point where we're well beyond level two now and now quite into level three, which is agents systems that can take actions. Okay, so we are Not at the beginning stages. We're now into like the middle stages of level three. I think we have more I think we have more maturity still to go with agents. to the point next year we'll be in full full mature level of agents by this time next year. That's my prediction. We're in we're in the middle phase now. La end of twenty twenty five, early this year, we were in the very beginning stages. Now it's maturing and it's like it should be pretty reliable. And again, next year we won't be talking so much about agents. We'll be talking about what's next. Which is level four, innovators, AI that can aid in invention. Okay, this is where it starts to get interesting. This is where AI starts to be able to help with the algorithm. Remember we talked about questioning the assumptions? We can't have innovators unless you can question the assumptions correctly. If AI can question the assumptions correctly, then we'll start to have innovations from AI. And that'll be really helpful. It won't necessarily be able to simplify the process, maybe not in the beginning, but as it gets smarter and is able to do more, it'll actually be able to operate more and do more of that algorithm for us. I still highly believe that there's gonna be a huge human element into it because there's lot of context and emotions and gut feelings that humans have that AI still doesn't have and will never have, especially when it comes to relational factors, trust factors that I think are distinctly human. But that's where they're going. We haven't seen this yet because AI struggles to question assumptions. It just moves forward. but we're of course seeing some early indicators of it. So maybe by this time next year we have mature agents and we start to get to the very beginning stages of AI that innovates. And then level five is organizations. AI that can do the work of an entire organization. Because if it can run agents and systems, if it can reason, if it can actually take innovative steps where it needs to, in just the right spots, then theoretically you can have whole teams of agents running a whole organization or a whole p sophisticated project. And that's where things start to get really interesting. I don't know what that's gonna look like. Nobody really does. We can theorize about it. but it's an idea, and I would say that this track has been on par so far. It's been two years since they published this and I'd say it's been very reliable. It's been a reliable p pe peak into the future, you could say. And I've covered this chart a few times on this show. But this innovation is starting to become clear what it even means. When we were back just dealing with reasoners, innovators were like, what does that even mean? Now it's like, it's because AI struggles to question the assumptions, even with great reasoning. It has a hard time taking a step back and looking at the bigger picture of things. It's a very human thing right now. But will it always be a human thing? And what starts changing as a result of AI being able to actually question assumptions? Maybe AI starts becoming more of our boss and we take orders from it. That'll be interesting. So all things to think about. But again, right now everyone's scrambling to learn how to build systems. You should learn how to scramble to build systems. But if you want to get a step ahead, start looking. Read the algorithm. It's a fantastic book. It's a four hour read, two hours on double time. really easy. But you need to start looking beyond creating systems and look for the actual thing holding back growth in your department. If you can start doing that now and get in the habit of seeing it and building a habit of questioning assumptions, doing things manually, and then helping people then create great automation systems with AI agents, you will be well ahead because everyone's creating systems No one's tr being trained on how to think critically about the whole process in the beginning. So if you do that, you will be well ahead of the pack. And of course it's good to keep in mind of where we're going and start thinking about how that might impact your job or your organization. Because someone's gonna be needy to be kicking off these agents. It might as well be you. Thanks for joining me on today's episode of the AI Driven Marketer.