The AI Shift Marketers Must Know: Claude Just Overtook ChatGPT

Well, hello, hello, hello, and welcome back to The Marketing Companion, The AI Marketing Companion actually. I'm Sandy Carter, I am your host. I've been working in AI for a long time and has previously been a CMO and done, I think, every absolute job in marketing that there possibly is. And I am so excited today because We found out that the, since the, for the first time since the AI race began, Anthropic has overtaken OpenAI in business adoption. And that is a really big, big deal if you think about what's happening in this space. Anthropic is now at 34%, OpenAI at 32%. So it's not like a huge difference, but it's really enough to matter. Now, this is according to Ramp AI Index, Anthropic leads OpenAI among US businesses. So for all my international folks that are listening, this does not apply to you. This is just showcasing how many first-time AI purchasing decisions now are going to Anthropic. But now we're talking to you guys. You guys are marketers. This story is bigger, not just because now Claude has overtaken ChatGPT, but because all of you are telling me that you're moving beyond a chatbot and into workflows. So as you're working, looking for something that can handle real work, not just answers, I think that really brings us into the latest release of Claude, and that's Opus 4.8. It is the new model that's out there. If you haven't tried it, When you go into Claude, you know how you can switch from models to model. Try Opus 4.8. And I think it has a really interesting, important feature that Sean and I are going to talk about today, which is dynamic workflows. So today I'm joined by Sean Reddy. He is the leader at Team Nebula, which is going up, up, up, up, up in growth. He was telling me about earlier, and we're going to discuss what this shift means for marketers, content creators, and business leaders. So welcome, Sean, to our show.

Thank you. Yeah, thanks for having me again.

And I have to say, I always want to thank our two sponsors. We have two sponsors for the AI Marketing Companion. One is SEMrush. And I know I always say this, but I use SEMrush at my business, both for AEO or that using all the, the, and, and, uh, agentic tools that are out there today, as well as SEO. It's awesome. So if you don't use it, try it out. And as well, I use Brevo for all of my customer communication. Communications. So after these podcasts, I send you guys a newsletter announcing that the podcast is live. I do all that through Brevo. I manage my, basically my CRM through Brevo. So if you haven't tried those two, please make sure that you go and check those out. Okay, Sean, are you ready to get into it?

Yeah, ready.

Okay, so I'm going to start out with the first question, just saying, you know, we, I just announced that OpenAI overtook— I'm sorry, that Claude Anthropic overtook OpenAI for the first time ever. Yeah. Why that matters to marketeers. What do you think?

Yeah, I think it's a real-life index of what businesses are buying. A year ago, companies were paying for like an assistant that answers questions, and now they're paying for systems that go do the work end-to-end. And that's the whole shift happening in one sentence. Like, the budget is moving from help me think to go execute. If you run marketing or you're in marketing and you're still treating AI as a smarter search bar, you're spending money on like the old category. And what everyone is spending money now on is full end-to-end assistance that can go take the actions you need them to take and you can actually trust them to do it properly.

Yeah, I think that's really important. You know, I did a podcast actually with my daughter, Sean, And one of the things that she was talking about is how she felt like her generation is outsourcing their thinking. And she's like, we really don't want to do it, but we're kind of lazy. Those were her words, not mine. So I think it's interesting that you talk that this is now enabling you to add that creativity and add your thought processes into the mix, which I think is super, super, super duper important. I will also add one more thing on that report that I find really interesting is that most businesses today are using both Claude and ChatGPT. So overall business adoption of AI today sits at about 50.6%. So if you add those two vendor shares together I gave you, it's obviously bigger. And so what that equals to is about a third of every company using AI is writing checks both to Anthropic and OpenAI at the same time. And that is because enterprises are using multimodal stacks. We do. I think you do as well. And the RAMP data proves that.

100%. Having the flexibility to switch to different models is so important for different use cases, whatever it may be, whether you want to run like a local stack due to security and compliance. The piece, and what we'll talk about a little later too, is the orchestration layer. And that has been what's been the hottest thing in AI lately in terms of projects coming out, open source projects, SaaS coming out. Everyone's focused on that because that makes all the difference. Yeah.

In fact, I just was at the Lambda event in Silicon Valley and I met a couple of companies that are focusing on the harness and the orchestration. And they were swamped. Like orchestration is hotter than hot, but we'll get into that here in a second as well. So now I think, you know, after this report, even before the report, everybody was talking about OPUS 4.8, this brand new model. If you use Claude, when you go into Claude, you have a choice of the model you want to use. 4.8 is one of those choices. And one of the big features that just came out is, I think, a very important one. But I don't know if a lot of people are looking at— I think a lot of people are missing out on it. Why don't you talk about that? What's that one feature that marketeers are missing?

Yeah, it's my new favorite feature and it's called Dynamic Workflows. And, you know, to break it down in the simplest way possible, like a prompt is one instruction and one output. You ask it, it answers. A dynamic workflow is the model deciding at runtime, you know, what steps the entire job needs, doing them and assembling the result. So better prompting makes, you know, one answer better, but a dynamic workflow handles the whole project. You know, if you want to think of it too in like real-world scenario with work and your company, a prompt is asking an employee like a single question. A workflow is handling, handling a manager and a project and trusting them to break it all down together so that it is as efficient as possible and gets you to your end result that you're looking for. And it'll also even delegate to— so say that you have an agent that's better at the marketing copy piece and there's an agent that's better at coding, it'll delegate to the right agents to go execute accordingly to get you to your North Star. So it's an entirely different way of getting the job done.

So let's think about it. So if we're going to do a prompt, we might say Hey, I'm thinking about doing a marketing campaign about this. What do you think? And Claude would come back with an answer. If I'm going to use the dynamic workflow, I might say, hey Claude, create me a campaign, pull together all the elements, and it would pull from the right agents and it would come back after a little while and give me the entire structure that I need for that campaign. Would that be correct? Would that be a good example?

Yeah. So how I would structure that prompt is I'd be like, hey, this, this is the campaign I'm looking to create. Please execute creating this campaign using dynamic workflows. And you'll see it kind of highlight the words dynamic workflows. It'll kind of like make it like rainbow colors. And that's how, you know, it's like locked on to that sequence. And what it would probably do is it would go deploy a research agent to gather all the context, deploy an agent that's focused on copy and actually, you know, writing out the, the project or the slides or whatever it may be. And then there'd be another agent maybe focused on the visuals and the graphic design, and it's doing all of that at the same time and orchestrating it in the right way to get a good result.

So yeah, but I do have to put dynamic workflow in the prompt. Is that how I— is that how I get it to work? Yeah.

Yes, yes. A lot of people think it's like a skill, like a bachelor skill, but it's not. You literally need to type it out like in your prompt. Yeah.

Yeah, I was, I was actually playing with it the other day because I missed that, Sean, when it first came out. And somebody who was watching me, they're like, oh, this is what you're doing wrong. So it was very helpful for them to do that for me. So, you know, some of the stuff we just talked about is a little bit complex. But for someone in marketing, tell me exactly what a dynamic workflow is and how is it different than just writing a better prompt? We kind of touched on that a little bit, but give us a couple more examples and that might make it, you know, real.

Yeah, absolutely. So, for example, if I were to write a more extensive prompt, I would have an agent normally coming back and forth. With me to gain more context and for me to give it more answers so it can work through its delegation process and its execution process. But that's a very hand-holding experience where I'm still having to be focused on micromanaging that agent. The beautiful thing about dynamic workflows is I feel like I moved from a project manager to the CEO where I'm like, okay, you go be the project manager, you go delegate to whatever employees you need to delegate to. I don't want to handhold you because I want to go do something else. That is the huge difference is I can literally actually like go do another task. Whereas like you can't really do that if you're not using dynamic workflows because you're having to make sure that it's not sitting there doing nothing waiting for you to respond. I think that's the biggest difference is like, I remember even there being some open source projects that came out where people were getting creative about notifying you when you had to respond to Claude. Right. You know what I mean? And so this is like getting around that where now you don't have to focus on that. You can trust that this is going to go execute end to end. You can get started on something else. It's a huge productivity hack.

And do you see, you see anybody else doing this? Like, could you do this with OpenAI today?

OpenAI has like different agentic protocols, but I have not experienced orchestrate agents with such little handholding on this level, if that makes sense. So it doesn't feel like it's as autonomous.

Yeah, I think that's right. I think that it can be a little bit harder to orchestrate. It's not that you can't do it. It's just it takes a lot more effort. So I think with Claude, you really, you really get the feel of what a dynamic workflow is and it just takes over. Takes over means automates a lot of things for you, which I think is really good. Okay, awesome. So let's go through one more real marketing example if we could. So if I start with a webinar recording or a podcast episode, hint, hint, like this one right now, how could Claude handle that entire workflow from content creation to distribution?

Yeah, that's a great marketing use case. So say you start with one podcast episode. The system can transcribe it. It can pull the 5 strongest ideas from the transcription, turn each into a LinkedIn post in your actual brand voice, not a generic AI voice. It'll cut the clipping, the clip moments, like the most significant moments that you can pull out from the transcript, draft potentially a newsletter, write, you know, a blog version of it. And queue everything for distribution all in your brand voice too, which is the key.

So it's taking the place of Riverside or another tool.

Yeah, I would say more so for the actual like writing version of the content. I would say for like video out of the box, Riverside is still better and structured for that. But Claude can even do it if you were to use some open source projects that I've seen that would allow you to clip. Video. So yeah, there's a lot of different options for executing content that you want to distribute. You can even draft your comment reply so engagement doesn't die as you start getting people, you know, responding to your content. And then one asset becomes, you know, 15 pieces across 4 channels. You can run 4 versions of this every week. The reason the new orchestration matters is reliability. And the system verifies each step instead of you babysitting every handoff.

Yeah, I think that's, I think that's really interesting, especially for marketeers who are, you know, super busy because a lot are doing more today, not less today. They're doing more today. So I think this is really important. Now, my last episode was on AI-generated content becoming AI slop. You know that word, right? Yeah. Um, and so I'm curious, do you think that these dynamic workflows now will make that problem worse? Or do you think it can actually improve the quality and the brand consistency?

Yeah, that's a great question. I think it, I think it could do both. I think which one you get is entirely up to you. For example, if you point a workflow at volume with no taste and no brand output, you get slop at scale, right? And that's the nightmare. That's not what you want. But if you encode your voice, your constraints, um, you know, potentially even human approval towards the end of its orchestration, the same system, you know, raises your floor and what you're able to accomplish. So slop was never a tooling problem. It's more a taste problem, is how I'd put it. And your workflow just amplifies whatever you feed it, you know. If you feed it garbage, you're going to get garbage at scale. If you feed it what you want to output in terms of the context, you're going to get that context at scale.

Well, you know what, Sean, that's really interesting to me. I don't know, did you just see that Nike ad that came out and they actually accidentally, they say accidentally used AI on it and it came out with the, you know, not this but that, you know, the typical AI format and they got crucified.

Yeah.

And then another company, they came out with an ad. And in the ad, of course, the ad was all about trust, why you should trust this brand, trust this brand, trust this brand. And they used AI to generate images and somebody had 6 fingers and they got cremated because it wasn't even about the slop. It was about how can I trust you when you've got actors, AI actors running around with 6 fingers? So I know you said that it's more about taste and brand. But I think it's even more than that, right? For marketeers looking at the content, if even if you use dynamic workflow, you still have to look at your work. You still don't— at least for me, I never take the first thing I get back. I always look at it. I always have it go back. I always have feedback on it. I'll even hand change things for it. Um, do you feel like enough people are doing that today, or do you feel like they're just defaulting to what comes out? Even with dynamic workflow, do you think people are now more compliment because it sounds really important, that they're just going to release stuff into the wild.

Yeah, I think a lot of people will do that. I think actually the people who will stand out in their vertical or industry are the people who are taking the extra time to, you know, utilize stuff like Dynamic Workflows, but, you know, comb through it, make sure that it's up to their standard, make the fixes that are necessary, give feedback. Refine it, whereas a lot of people are just like, oh, this is so quick and easy. They almost get a rush out of just like doing it, not having to do anything with it themselves, and just putting it out there. And that content is, I think, over time is going to be easier for us humans to just recognize, oh, AI slop, or oh, AI, they didn't touch it, there's not a human involved with this. And I think that's going to become less valuable and actually have the reverse effect in terms of of how people receive it.

Yeah, I used to have on my LinkedIn page, you know, human created because I would write my own post. I then had AI edit it and then people were like, oh, I see this editing. So it's not— so I didn't leave it there because yes, it's true. I used AI to edit content. I'm horrible at grammar. I'm horrible at spelling. But I still— all the thought, all the content is mine. It's my original. Content, my original thought. And I do think, you know, even with the dynamic workflow, when I've used it, it comes out high quality. But a lot of times I've had to run it multiple times. Are you seeing the same thing? Like, you still have to have that human in the loop, right?

Yeah. What I found is if I go back and forth with Claude and don't include dynamic workflows right away, so I'm able to like build out a conversation with Claude first to give it context as well as ask for feedback and we can like go build more of context leading into the dynamic workflow, I tend to get a better end result and I don't really have to invoke another dynamic workflow if I do that before.

Yeah, it's quite interesting. So everybody in marketing out there, you know, this is not replacing you. I get a lot of questions, will AI replace me? I don't think so. I think the good quality job, the creativity, coming up with that big idea is still with you and reviewing all the stuff. You know, I think about it kind of as having an intern or someone who's a junior working with me. I would never just push their content out. I would always still do the review of it, help improve it, you know, teach as well. But that would always be the way that I would orchestrate it just so that things are gonna become even better.

Yeah, absolutely.

Now, we did touch on orchestration. Do you want to talk just for a second, Sean, about like what is orchestration to you and why is it so important?

Yeah. Orchestration allows me to have the largest footprint with AI without me having to tell every individual agent or sub-agent exactly what they need to do. When I'm able to spin up more agents to go execute a task, I feel like they're able to think things through 2 to 3 more steps, which is like really important for getting better quality in your end result. For example, I used to, when I would build presentations, I would get the presentation in Google Slides and the formatting would be all messed up. Like the visuals, like there would be some visual components overlapping the words. Some of the words wouldn't centered. Like, the, the end result was just like, oh my gosh, I can't use this. I just have to use it as a PDF because that's the only way it would come out properly. Now I'll, you know, export it into Google Slides and everything will look perfect. And that is due to more sub-agents being deployed and some agents being more specialized in the visual component and being able to focus on that. Whereas when there were less sub-agents, it was just so focused on getting the in result out that it wasn't looking at as many minute details. And I think, you know, when you're working on professional things, you've got to be looking at those details. That's what a human would be doing. And so it's making all the difference having a higher level of details in your orchestration layer.

And just for everybody, an orchestration layer is really easy. It sounds very complicated, but it's just the way that you direct agents to work together. So these companies who are developing orchestration software, it's just saying, oh, this goes before that, or if this condition applies, then that goes first, or this goes first. So it's really like an orchestra of a band is orchestrating everything to work well together. And this is why it's becoming so popular nowadays, is because so many companies and so many marketing organizations now have multiple agents that they have to elevate and orchestrate. It's also one of the top 20 new jobs, is a job called AI orchestrator, to help figure out how all these things are put together. So if you're looking at that, I think that's— I think it's a fascinating new field and new area to look at as well. Um, speaking about jobs, Sean, what marketing jobs or What skills do you think become more valuable as AI moves from answering a question to managing a workflow?

Yeah, I mean, I think this is a great question because I think a lot of people have fear around AI replacing their job. And, you know, I just want to say the moat for you as marketers was never the model and utilizing some specific model. Everyone had access to the same model. The companies that pull ahead in terms of how they use AI are companies that are leaning on systems and proprietary context around them. I think 2 years from now, people will look back and realize the winners weren't the ones who picked the smartest AI. They're the ones who redesigned how their team and how the AI work together. And so what I'm saying is, I think how you go about utilizing AI, like putting more effort into the context before you deploy dynamic workflows, still looking over your work when it gives you the end result and tweaking it to your own personal touch, making sure it has your brand voice. Those are all things where you are still super valuable. We need you to work with the AI to execute in that fashion. The AI can't just go and do those things on its own, and that is where you will always be valuable, especially when you can put your personal touch on it. The AI can do, you know, whatever you tell it to do. You are kind of the central piece of orchestrating the AI to do it in the right way and do the right thing. So yeah, I think, you know, people have to kind of shift their mindset around their value as a marketer and deploy their value in a certain way alongside AI to really get the most out of it.

Yeah, I think that AI is always stronger when it's augmented. That's why I write the book AI First Human Always. Because I do think the value comes from augmenting what we can do, not replacing what we can do as well. So I'm a CMO listening today. Sean, what do you think is one workflow I should implement in the next 90 days that could immediately either impact my team's productivity or increase my customer experience?

Yeah, that's a great question. I'm super glad you asked me that because there's actually something that I'm working on right now that I think a lot of marketers could utilize and a lot of people are already doing. You know, I've seen a lot of it on X specifically, which is building content repurposing engines. A lot of content going out on X, even if it is valuable, if somebody has a content repurposing engine, they're still sourcing, you know, what is performing well, why it's performing well, and then putting their own personal touch on it and reshaping it for their X account. I'm seeing that like everywhere. And you can also take X posts that are performing really well and reshape it for other distribution channels too, whether you want to convert it to a newsletter, whatever it may be. But I would build a content repurposing engine, take one of your best pillar assets every week, your best, your best tweet, convert it to a good topic for a webinar, podcast, a keynote, whatever it may be, you can turn into a full multi-channel distribution run. I believe it's the highest ROI because you're going off numbers of something you know has already performed well in the space. So it's the lowest risk, I would say, in terms of a place to start. And the output is immediate. It doesn't take, you know, weeks to put together. It's— especially if you invoke dynamic workflows, it's a pretty quick process and it's pretty quick. Pretty cheap as well. I mean, in terms of the cost, you don't need to tend to deploy multiple dynamic workflows. So your token usage should be manageable for building a system like this. Yeah, I would say that's a great use case that I'm seeing everywhere.

You know, the reason I like that is so if you're a CMO and you're a little afraid of AI slop, you could take, let's say, let's take a LinkedIn post that's doing really well that your team wrote. Okay, that post is doing great, human written. Then you could run the dynamic workflow in Claude and ask Claude to help you repurpose that content. Maybe you want to put it on Instagram. Maybe you want to put it on X. Maybe you want to have it on WhatsApp. Well, maybe a WhatsApp campaign too, right? Going direct to your customers. And that's a great thing to try because then you're not producing more AI slop. What you're doing is you're taking your original content and you're using this as a content engine to be able to take that everywhere. I have to tell you, I was with a, a CMO who was telling me that the amount of time they spend just resizing pictures, you know, for the different formats, for TikTok, for Snap, for Instagram, for LinkedIn. This is a great way to try that out and really increase your productivity, right?

100%. Yeah, I will say OpenAI still is, you know, way further ahead when it comes to ImageGen. Yeah. But other than that, in terms of just like edits, yeah, Claude still does great at that.

Okay. Fast forward 2 years. What's the biggest thing that marketeers are going to look back on and go, oh my gosh, I completely underestimated that?

Yeah, that's a great question. I think, I think the biggest thing is there are some tools, especially in that you would normally employ someone that you think has to be a creative for, that you can use to assist and enhance a creative process. So for example, video editing. A lot of people pay, you know, really expensive video editors to get their video content done. But I have actually seen like open source projects that automate what a vlog's format should be like. And you can just drop all your footage in. Claude Code will go in and edit and work with Final Cut Pro to output a full edited video for you. So there's processes like that. Same with graphic design where people are like, oh, I don't want AI slop when it comes to creative processes. But I think it's all about utilizing the most innovative AI processes to assist your creativity and deploy your creativity at scale. It's, it's like a productivity hack is how I see it. It's like the more you can utilize AI, the more that you can deploy whatever you have that's unique to the table on a larger scale.. If you're not doing that, if you feel like AI is dimming your light, then you're using it wrong. And I think that is something where in a few years when people kind of get past like that frustration that they feel in terms of viewing it like it's replacing them, they'll be like, oh wow, like I was being held back so much by not trying some of these latest tools and, and, you know, strategies because yeah, they're not meant to replace you, they're meant to enhance you. And I think that it's going to just take time for people to really see.

That's interesting that you think it's going to be on the video editing and the content, that sort of thing. I think that's interesting. Okay. We're going to do a lightning round. So you're just going to give me quick answers. One marketing task every leader should automate.

Yeah, I would say definitely any short-form content strategy. So like, like scripting what the strategy is going to be like. Being able to understand the content that the strategy is going to be focused on, and then being able to have a strategy for redistribute— redistributing that. I think that should all be an automation pipeline.

One task humans will always own.

I think the emotion side, whenever we're portraying anything that should be pushing through some sort of ethos in copy, I don't think AI can ever replicate that on the same level a human can because human to human will always be different for that side of things.

I like that, the emotion side. We have another podcast coming up on that too. So, okay, most overrated AI trend.

I'd say the most overrated AI trend. Oh, is AI trading. I don't know if you saw that on X. It went super viral where everyone was posting, giving their OpenClaw, their crypto wallets, and it was, it was so overrated. Yeah, I'm not saying that there wasn't one person who made money, but it was just overrated.

Yeah. An underrated AI trend?

Underrated would definitely be utilizing OpenClaw for running automation workflows. A lot of people feel like it was unreliable, but it was all about the way that you set set it up and structured it.

Do you know I switched to Hermes now?

I know a lot of people do. Yeah, yeah, I'm a loyal Open Claw person, but— oh, you are? Yeah.

Okay, Claude or ChatGPT?

Claude.

Okay, good. One word that describes the future of marketing.

Oh, one word, uh, That's, that's a tough one. One word that describes the future of— I would say orchestration.

I would say promising. I think marketing is promising. I think that there are so many tools we're going to be using. I'm working with one marketing company right now and they just keep adding agents, like they're just growing. Their team has high morale. I think it's promising. I think once people figure it out, This is going to be great.

Yeah, I think it is super promising. There's, there's so much more that we can do as marketers with these tools at our disposal. It's wild.

Well, thank you, Sean, so much for being on here and talking to us about dynamic workflow and about the battle between OpenAI and Claude. We really appreciate it. Why don't you tell our listeners and our viewers a little bit about how to reach you?

Yeah. So you can check out my company at teamnebula.ai. We're an official enterprise partner of Anthropic now. We are focused on a lot of— Congratulations. Thank you. Yeah, it's super exciting because as you can tell, I'm a big Anthropic fan. We've been utilizing Claude for a lot of projects and we finally got to that point where we're now partners working on a number of government contracts, specifically focused on taking data, working with our partner to do data warehousing and then use that to fuel AI agent infrastructure. Like we talked about, the context layer, super important. Orchestration layer, super important. And then when you look at government contracts, it's all about security and compliance. And so how you structure that is just like key. So that's what I'm focused on.

That's why— teamnebula.ai.

Yeah, teamnebula.ai is our company. You can check us out. And then on socials, we're just Team Nebula AI.

Okay, awesome. Well, thank you, Sean. And hopefully everybody took away one major takeaway from the conversation that the future really isn't about using the smartest AI model. It's building the smartest workflow around it. And it really doesn't matter if you're a Claude user or ChatGPT user or something else entirely. Marketing teams that are pulling ahead are redesigning how humans and AI work together. And I think that is the real story from today's session is how marketing and humans will work together to take this to the next next, next level. Thanks again to SEMrush and Brevo. Thank you, Sean, for being our guest, and thank you to everyone who listened today. Thank you, and we'll see you soon again on the AI Marketing Companion.