AI in Marketing: What Every B2B Founder Actually Needs to Know
- Richard McClurg
- 7 hours ago
- 8 min read
In 1995, if your company had a website, it was novel enough to be listed in a printed directory of websites—an actual book, like a phone book, but for URLs.
I’m not making that up. (I was rather chuffed to get my company’s website listed).
Nobody knew what to do with it. Nobody knew what it would become. There was a lot of noise, a lot of “you have to get on this,” and a lot of companies building websites because everyone else was, with no real idea of why. (I do remember telling clients in 1999, fresh out of MBA school, that if they weren’t online, they wouldn’t be around in five years). Perhaps a tad over the top.
The arrival of ChatGPT in November 2022 felt exactly the same.
Three-plus years in, most of the founders and CEOs I talk to are somewhere between genuinely curious and quietly overwhelmed. Every vendor claims their product is now “AI-enabled” or “AI-powered.” Your team is probably using AI tools that you or your IT person (if you have one) doesn’t know about. There’s a cobbled-together AI policy somewhere that nobody reads. And the pace of change makes it hard to know whether you’re behind, ahead, or just confused.
You’re not alone in any of that. So let me give you a clearer picture of what’s actually happening, what it means for your marketing, and what to do about it.
The Noise Is Real. So Is the Risk.
Seven of the top ten companies by market cap on the S&P 500 have a significant AI play. Agentic AI—tools that can autonomously control software and take action on your behalf—is moving from concept to reality faster than most people expected. People are also talking about AI knowledge blocks (such as market, company, and product information), skills (think of a toolbox of capabilities for taking specific actions), and teammates (collaborative partners that take initiative and ask for help only when needed). In some developer circles, natural language is the new programming language.
At the beginning of 2025, the best models scored in the low single digits on Humanity’s Last Exam (designed to test AI at the frontier of human knowledge). By the end of the year, Gemini 3 Pro achieved 37.5% accuracy. That’s a massive (scary?) leap in 12 months.
This isn’t hype. But a lot of what surrounds it is.
Here’s the thing that keeps me cautious: marketers have a long history of exploiting any channel until it’s broken. Junk fax. Robocalls and text spam. Email spam. LinkedIn direct message pitch-slaps. Yeah, it’s part of the reason why (nearly) everyone hates marketers. The same pattern is starting with AI-generated content, where people are gaming AI answer engines—publishing misleading or outright false information presented as fact, just to get visibility. It erodes trust across the board, for everyone. Don’t be part of that.
AI advisor Nicole Leffer has been tracking this closely. When someone shares something impressive they did with AI and says, "you can do it too," you usually don't see the years of foundational knowledge they used to do it safely. You don't see the risk assessment. You don't see the guardrails. Most people jumping in right now don't know what they don't know. And some are putting their companies at genuine risk by sharing proprietary data, violating their own AI policies, and giving autonomous tools access to systems without understanding the consequences.
FOMO is not a strategy. Full stop.
Where Does Your Team Actually Sit?
Before you make any decisions about AI and marketing, you need an honest answer to this question.
Christopher Penn of Trust Insights has a useful way of thinking about AI skill levels—more specific than the usual “beginner/advanced” labels that tell you nothing. He defines five levels:
Beginning: Just starting with tools like ChatGPT, or not yet started
Intermediate: Using AI consistently and generating usable outputs
Advanced: Building custom mini-apps such as ChatGPT GPTs, Gemini Gems, Claude Projects
Expert: Connecting AI into workflows using tools like n8n, Make, or Zapier
Master: Writing and deploying AI-enabled code in high-code environments
Just because someone on your team has used ChatGPT doesn’t make them an AI expert—much the same way browsing the web doesn’t make you an internet strategist. Most marketing teams, if they’re being honest, are at the intermediate level. That’s fine. But the gap between “we use AI” and “we actually know what we’re doing with AI” matters when you’re making decisions about tools, hiring, and risk.
Where is your team? Where are you?
I’ll be transparent about my own level: using Christopher’s framework, I’m at the advanced level and experimenting at the expert level. I started with prompting, put real time (including many training hours) into getting good at it, then built a prompt library, knowledge blocks, custom GPTs and Claude Projects. I’ve dramatically reduced the time spent on repetitive, research-heavy activities, shifting my focus to higher-value work. I can also use AI to identify granular patterns and synthesize vast amounts of information from many sources. It really is a “force multiplier” for human capacity in research, competitive comparisons, customer call analysis, content ideation, content analysis, and more. I’ve built a digital twin trained on all my content, articles, and frameworks. I’ve built a Claude skill that applies my brand style guide consistently across everything I produce.
I’m exploring agentic capabilities, but carefully. For example, I give AI access to specific folders and “human-in-the-loop” controlled access to trusted websites. Not my email. Not yet. (Maybe never?) And for sensitive data, I use open-source AI models running locally on my Mac, not in the cloud. And I don’t use the free plans of AI tools for privacy reasons and data training practices. That progression didn’t happen overnight, and it didn’t start with the most complex tools.
Understand Where AI Is Heading
Christopher also describes four stages of how AI is being used—a progression that explains why “agentic AI” keeps coming up and what it actually means for your team.
Done by you: You paste a prompt in, copy the result out. You’re the middleware.
Done with you: Tools with some automation built in—GPTs, Gems, Claude Projects—that still need significant guidance from you.
Done for you: You provide the idea and context; the AI executes and returns a result. Powerful, but only works well when inputs and guardrails are well-designed.
Done without you: The emerging fourth stage, where AI reasoning and judgment take on more, and you progressively hand off more.
Most companies today are at the 'done by you' or 'with you' stage. The ‘done for you’ stage is now more accessible (especially with Claude Cowork)—but I’d say you still need to know what you’re doing to get the best use of it. The ‘done without you’ stage is certainly coming fast.
Build Intelligently, Not Reactively
Liza Adams of GrowthPath Partners describes how most organizations evolve their use of AI: starting with AI as a tool, graduating to building AI teammates, and eventually integrating them into workflows. Most organizations are at the tool or early teammate level, using AI for individual tasks but not connecting those tasks into anything coherent.
She recently cited McKinsey’s 2025 State of AI report, which found that workflow redesign drives the biggest impact from generative AI, yet only 21% of organizations have actually done it. Most are bolting AI onto existing processes and wondering why the gains are incremental.
The insight for your marketing team isn’t “get more AI tools.” It’s “connect the ones you have.” A standalone AI teammate for competitive research, one for audience/persona, one for battle cards—those are only as valuable as the humans who understand them and the workflows that connect them.
As Liza puts it, weak links in the chain make problems worse, not better. If you automate a broken process, you just get a faster broken process.
Don’t Skip the Change Management
None of this works without it.
Katie Robbert, CEO of Trust Insights, applies a 5P framework to planning and building technology implementation (including AI adoption): Purpose, People, Process, Platform, and Performance. The mistake most companies make is focusing on the platform—the shiny tools—while ignoring the other four. What are you trying to accomplish? Who needs to be involved? How are we going to accomplish your goal? How do you measure whether it’s working?
AI policies that nobody reads won’t protect you. Brief your team properly. Be clear about what can and can’t go into AI tools—confidential data, customer information, and proprietary ideas don’t belong in cloud-based AI tools unless you’ve verified how that data is handled. Build in human review before AI-assisted outputs go anywhere important.
The Overwhelm Is Real—And It’s Okay
Nicole recently posted something that resonated with many people: she’s overwhelmed. Not just by AI—by the pace of everything. And she’s spent years deeper in this than most.
Her approach was practical: separate what you can control from what you can’t. You can’t control the pace of AI advancement. You can control which changes are actually relevant to your business, how much time you carve out to learn, and what boundaries you set around what AI tools can access. That’s enough to start. You don’t need to be an expert. You need to be curious, deliberate, and honest about where you are.
What This Means for Your Marketing Team
Here’s the practical question for you as a founder or CEO: What kind of marketer do you need?
Not an AI expert who happens to know a bit about marketing. A marketing expert—and I mean the full sense of marketing, not just the promotions piece of it—who is genuinely curious about AI, willing to experiment, and capable of knowing when the AI is going off the rails. Rich context and domain expertise are what make AI output valuable—you need someone who can spot when something sounds smart but is actually wrong. And AI is very good at sounding smart (as my recent interaction with Perplexity demonstrates).
The model worth building toward: a small, AI-savvy human team supported by AI teammates designed for specific workflows. Think of the human team as your core—maybe it’s a product marketer, a content person, perhaps a field marketing/events person—each working alongside AI teammates that handle research, competitive analysis, campaign performance, content drafts, and more. The humans bring context, strategy, judgment, and critical thinking. The AI brings speed, scale, and the ability to compress days of work into hours.
This isn’t theoretical. Nimble B2B companies are already competing this way against organizations ten times their size.
Crawl, walk, run. Start small. Assess where you are, identify the highest-value use cases, and build capability before you build complexity. Document what works. Share workflows and automations across your team. Create an environment where experimenting and failing are safe—because you’ll need both before you find what actually works.
The Foundation Still Comes First
One caution before you go building AI teammates and workflows: AI amplifies what’s already there.
If your positioning is clear, AI helps you communicate it faster and more consistently. If your positioning is fuzzy, AI produces more fuzz—faster. If your messaging is solid, AI can scale it across channels. If it’s muddled, you’ll just muddle at speed.
Real MARKETing still starts with clarity—on who you’re for, what makes you meaningfully different, and how to explain it in a way your market actually understands. Then strategy to figure out how best to reach your ideal customers. Then execution. That order matters. What AI changes is how quickly you can execute once the first two are sorted.
The tools will keep evolving. The foundations won’t.
Start with what you can control.
Not sure where your marketing and AI stack up? Let's find out.
