AI Reality Check: What CMOs Are Actually Building - with Nicole Leffer

LinkedIn would have you believe your competitors are running fully autonomous AI teams. They're not.

The real picture of AI adoption in B2B marketing is far more varied — and far more instructive. Most marketing organizations are somewhere in the middle: using AI through a chat interface daily, but nowhere near the agentic workflows the feed suggests are already table stakes. Understanding where you actually stand, and where you intentionally want to go, turns out to be the strategic question most CMOs haven't stopped to ask.

Nicole Leffer is an AI consultant who works exclusively with B2B marketing leaders and their teams — from Series B companies through Fortune 50 brands. She spends her days inside marketing organizations helping them build AI marketing strategies that are grounded, intentional, and built to last. This is her second appearance on Growth Activated, and a lot has changed in a year.

In this episode, Mandy Hornaday and Nicole dig into the real state of AI adoption, why intentionality is the piece most CMOs are missing, the security risks of building without IT support, and the one AI strategy decision Nicole believes will separate teams built for the long term from the ones who will have to rebuild.

What we cover:

• The real state of AI adoption across B2B marketing — not LinkedIn's version, the actual one
• Why intentionality is the most important thing most marketing leaders are skipping right now
• The security risk of expecting marketers to build like engineers — and what smart orgs do differently
• Build vs. buy: when your core AI tools are enough and when they're not
• How to build a platform-agnostic AI marketing strategy that won't lock you in

Chapter Markers:

• (00:00) Welcome & Real AI State
• (09:58) The Intentionality Imperative
• (22:21) Building Without Silos
• (31:44) A Framework for CMO Action
• (41:54) Build vs. Buy & Final Advice

Connect with Nicole Leffer:

LinkedIn: https://www.linkedin.com/in/nicoleleffer/

If this episode challenged the way you're thinking about AI strategy, share it with a marketing leader who needs to hear it. And follow Growth Activated on Spotify or Apple Podcasts so you never miss an episode.

Growth Activated is produced by Mandy Hornaday.

Lead Like a CMO - Group Coaching Lab: Join the Waitlist

Let’s Keep the Conversation Going!
Loved this episode? Connect with me for more insights on B2B marketing leadership and strategies to grow your business.

🌐 Visit my website: growthactivated.com
🔗 Connect with me on LinkedIn: Mandy Walker
🔗 Get Your Free Marketing Planning Guide Today!

Don’t forget to subscribe to Growth Activated and share this episode with fellow marketing leaders. Let’s activate growth—together!

Transcript

Mandy Hornaday:

Welcome to Growth Activated. I'm Mandy Hornaday, your host with 15 years of experience leading marketing teams ranging from small startups to large service organizations. I've built high-performing teams of all sizes and have seen firsthand how fast the landscape is evolving, making marketing leadership more complex than ever. Today, I help marketing leaders elevate their strategies, lead with confidence, and build careers they love.

If you're ready to drive impact and unlock growth for yourself and your company, you're in the right place. Let's get started.

Mandy Hornaday:

If you've spent any time on LinkedIn lately, you might be thinking that most marketing teams are already running autonomous AI agents, orchestrating entire workflows while sipping coffee and going for walks. But the reality? The vast majority of companies are still figuring out how to get their teams to use AI consistently, let alone build anything autonomous.

There is no average right now. Some organizations are barely getting started. A very small number are doing genuinely advanced things. And a huge swath is somewhere in the middle — using AI daily through a chat interface, but nowhere near what the feed would suggest is table stakes. Understanding where you actually are and where you intentionally want to go turns out to be the most important question most marketing leaders are not asking.

My guest today is back for her second Growth Activated conversation, which feels right because a lot has changed in the year since we last spoke. Nicole Leffer is an AI consultant who works exclusively with B2B marketing leaders and their teams — from Series B companies all the way through Fortune 50 household names. She spends her days inside these organizations, helping them build AI strategies that are grounded, intentional, and built to last — not just built to impress.

In today's conversation, we get into what the current state of AI adoption actually looks like across the B2B marketing organizations Nicole works with, why intentionality is the thing that most leaders are skipping, and the one strategic decision she believes will separate the teams who are set up for the long term from the ones who will have to rebuild. If you're a CMO or marketing leader trying to figure out where to focus right now — not what's flashy, but what's actually going to matter — this one is worth your full attention. Let's get into it.

Mandy Hornaday:

Hey Nicole, welcome back to Growth Activated. I'm so excited to have you here today.

Nicole Leffer:

Thank you so much for having me. I always love a conversation with you, Mandy. Super excited for today.

Mandy Hornaday:

Me too. And I have to say, you are my first repeat guest on Growth Activated. I cannot believe it's been a year since we last spoke. Totally appropriate for you to be the first one to come on again.

Nicole Leffer:

That's so nice. Yeah — a year in human time is like 25 years in AI time. So absolutely.

Mandy Hornaday:

It's wild when I reflect back on our conversations. We talked so much about workflow automation and that was really helpful for me. But before we even get into some of my more tactical questions, I would love for you to just level set with us from a CMO perspective. I know you work with so many different B2B marketing teams. What would you say is an accurate current state of AI in marketing? Because if you just look at LinkedIn, the perspective feels a bit out of touch.

Nicole Leffer:

LinkedIn is definitely out of touch. Wildly varied — there is no average right now. We have companies who are literally just starting to tiptoe in, thinking, "Maybe we should get ChatGPT or Copilot," or they have Copilot and they're trying to figure out how to use it. They're barely getting started — where a lot of people were two or three years ago.

And then on the other extreme, I know a marketing leader whose team wasn't adopting AI. She couldn't deal with it. She fired all of them, rehired people who are deep in AI, and they are building truly autonomous, agentic workflows connected to all their data systems. They are all agentic orchestrators. But that is the 0.1% of the 0.1% — a very small fraction of reality. And to be clear, very few companies should be doing anything even remotely close to that.

She wasn't doing layoffs to cut jobs for AI to do the work. She replaced people with people who take AI seriously — one to one. I'm not seeing layoffs driven by AI, except for skills-related ones. Just switching out the skills.

The thing that's far more common is somewhere in the middle. Maybe they've started — they have ChatGPT, Claude, Copilot, or Gemini as their core tool. They're putting together some continuous workflows. Everybody's using it daily, but really through a chat interface. That's still kind of the average. But I honestly see just as many companies barely starting, where all they have access to is Copilot and they've never really used it. It is almost shocking how wide the variety is.

Mandy Hornaday:

Nicole, just to level set — what is your ICP when you think about which types of companies you're working with most?

Nicole Leffer:

I'm working with B2B companies — anywhere from Series A startups, though usually by the time they're working with me they're at least Series B or C, all the way through Fortune 50 household names. Usually there's at least 10 or 15 people on the marketing team, sometimes 150 or more. Almost all B2B, largely tech — the tech ones tend to be further ahead. But I also work with other B2B enterprise companies, not just SaaS.

Mandy Hornaday:

So I'm a bit stuck on that 0.1% person you described. From my perspective, the pressure is increasing because what we know is possible is so drastically different from where we're at today. The gap feels like it's growing between what we're learning conceptually and where we're actually at in practice. How are the CMOs you're working with thinking about bridging that gap? How are you coaching them to approach it?

Nicole Leffer:

I'll speak more to how I'm coaching them, to be careful about client confidentiality. The big thing I'm really starting to talk to marketing leaders about is intentionality. What is your end goal? What are you actually trying to do?

Just because certain things might be possible, or theoretically the future — is that even the future you are trying to build toward? Before you run so fast in a specific direction, decide if it's even the direction you and your company want to be running.

I really hope people are not building with an end goal of eliminating all the humans. But I see some people on the cutting edge who are heading in that direction, and I wonder if that's even their intention. We need to be building our AI strategies with intentionality around what we are ultimately trying to accomplish. Because until you understand what you're ultimately trying to build, you can't really build it.

Think about it this way: if you're trying to build a structure and you just start digging a foundation and throwing up walls without knowing what you're building — are you building a high-rise or a one-bedroom shack? What exactly do you want to get out of this technology?

Are you wanting something that works seamlessly across your entire org? Are you trying to build just for your marketing team? Are you trying to help your team be as efficient as possible, or keep the same team size and do the work of 10 or 20 people with three? If you don't slow down and ask yourself that, you could be building completely the wrong things on completely the wrong foundation.

I think we're going to start seeing a lot of crumbling in marketing because people have no idea what they're building — but they're building it anyway. Before you go too far, you need to think about that. Everybody is running so fast that they're not stopping to ask: what am I even running toward?

Do you really want to be the one eliminating the humans? Is that actually the best long-term strategy? I don't think it is. I think the best strategy is to keep your headcount where it is and massively increase what your company can accomplish — expanding your goals and capacity. Those are the companies that are going to win.

Mandy Hornaday:

This feels like an age-old problem for marketing teams. We are activity-based. We just run. We don't necessarily stop to think about the strategy of what we're doing and why. AI amplifies a lot of that chaos if you've never built strong operational processes and systems at the root.

Nicole Leffer:

I see so many people building these wild autonomous workflows and I'm like, what are you even actually doing? It looks cool. It's shiny on LinkedIn. But you have 850 different dashboards you've coded — are you really using all of them? What's the point? And if you don't know the point, you're just wasting your time.

Mandy Hornaday:

Absolutely. And another problem we've always had in marketing is activation and adoption. I see the same things — there's competitive research delivered to your inbox every week, but who's using it? Where is it influencing your other processes? Is sales actually consuming what you're giving them? It's just more noise. Are we overwhelming ourselves?

Nicole Leffer:

Right. And one of the benefits of this technology should be that it gives us space back — thinking time, maybe a little more balance in our lives. Maybe marketing leaders don't have to work an insane number of hours.

But instead I'm seeing people give up their nights, their weekends, their entire lives to build all of this AI stuff — to what end? To automate yourself out of a job? If you have a team of 80 people, you don't have to give up your whole life to build these AI things.

I'm in a CMO Slack community, and someone asked, "What do you guys do while the agent is running?" And I commented: go for a walk. A few others agreed. But most people's mindset is that you have to keep context-switching constantly. No. If the hard work is being done for you and you're going to come back and review it — go enjoy your life during that time.

We get to decide what we want this technology to be. It is not inevitable. Leaders are the ones who are going to set those norms. Maybe we could build a future where we actually breathe a little more and enjoy our lives. I think it's something we should all be thinking about.

Mandy Hornaday:

That's interesting. I've considered going back in-house, and for me the dream is running a flat, lean team of really smart people powered by AI. I used to run a team of 25 and honestly didn't love that. I love the actual work and collaborating with smart people. But are CMOs bumping up against what CEOs and CFOs think they need to be building? You could say, "I want to keep my whole team but 5X or 10X our outcomes" — but are finance leaders on board with that?

Nicole Leffer:

That obviously depends on the company. But honestly, what I'm seeing more than anything is that everybody feels like they're supposed to embrace AI and nobody has a clue what they actually mean by that. CEOs, CFOs, CMOs — there's just this sense of: run as fast as you can at embracing AI.

But right now you can absolutely make the case: we're going to use this to multiply our revenue generation, what we can accomplish — and we're not just going to do the same with fewer people, because you still need that human expertise. You still need that human curation. You still need the person who knows what good is. Maybe more than ever.

Because there's a gap now between people who sound productive and people who are actually creating quality work. A lot of what you see on LinkedIn — if you look at the actual output, it's not good. Just because it's on LinkedIn doesn't mean it's actually good.

Wherever you fall on the adoption scale, you can be scaling garbage very quickly. What CMOs need to communicate to leadership is: we need these humans so that what we're scaling is good, quality, effective, resonant marketing — not just marketing for the sake of marketing. These tools are brilliant and capable of incredible things. But they need someone steering and reviewing them.

So few companies are putting together the guardrails for not just what goes in, but how to guide the AI on what good actually looks like. That's where the human is so important. If your team won't do that, you need people who will.

Mandy Hornaday:

And one thing I don't see a lot of marketing leaders thinking about — when we talk about automating our work — is the human element of alignment and collaboration. Bringing the rest of the org along with you. How are speed and alignment working together? Or are we just going fast in marketing and leaving everyone else behind?

Nicole Leffer:

It depends so much on org size. The bigger the org, the slower they're moving — largely because they're doing a lot of that internal alignment. Two years ago I might have said those big companies are moving too slow. Now I'm starting to think some smaller companies are moving too fast, and that's going to catch them off guard. I've actually shifted my perspective on that.

You do need to make sure what you're building isn't in a silo. When you're moving so fast, you run the risk of not talking outside your marketing org — not communicating with sales, product, CS, and the rest of your organization.

Marketing in many ways is ahead on AI, but if you consider product and engineering at tech companies, they're as deep as or deeper than marketing. So they might be building one thing with AI while you're building something completely counter in your silo — and you'll have issues. And if sales isn't on board with what you're building, just because you had the idea doesn't mean it's a good one. It has to fit in the whole puzzle.

Mandy Hornaday:

Out of curiosity — are you finding that marketing teams are leaning on their tech teams to help build? Or are the orgs that are winning owning the end-to-end build themselves?

Nicole Leffer:

I'm not seeing a lot of companies where product engineers are helping build marketing AI stuff. I've seen a couple — they're doing some cool things because they're using those resources well. But mostly I'm seeing marketing lean on IT-type tech support for the more involved pieces.

And there's a real risk here. As a CMO — unless you come from an engineering background and genuinely understand cybersecurity and data security — you and your team are probably not equipped to build the way some people think you should be. This is one of the things people are getting really confused about with AI: thinking marketers are supposed to suddenly become engineers and cybersecurity experts. That is risky. We're going to see some real problems because people are building so fast without considering that.

The smart orgs are working very closely with IT security. They are not setting their marketing teams up to use Claude Code inside their computer terminal and build their own custom connections to databases. You would think, based on LinkedIn, that's what everybody is doing. But your team probably does not have the knowledge to securely build those data connections in a way that isn't putting your internal data at serious risk.

I think we got really spoiled by the fact that AI tools seem to know everything about everything — and we forgot that humans don't. You can ask ChatGPT or Claude about basically any topic and it's an expert. But humans are not. You can't just assign humans to do expert-level things outside of their own experience and background. That's a mistake I'm seeing a lot of marketing leaders make.

Mandy Hornaday:

I did my own little Claude Code hackathon weekend a couple of weekends ago — but only for my own personal stuff. I wouldn't do it for clients because I know there are things I don't know, and that does worry me. Even the question of having it rewrite or update live files — I don't know enough to know whether that's safe or secure, so I haven't taken that step.

Nicole Leffer:

I want to be clear — I'm not saying nobody should be using Claude Code. It is a developer and engineer tool, and there is some really cool stuff you can do with it. There are definitely marketers who are fully ready to use it. But when I hear leaders say "I'm going to get my whole team set up on this" — is your whole team made up of engineers? Do they all have the understanding of what comes with it?

They're doing autonomous stuff. They're building connections to their tools. They're giving it internet access. And they're stacking things together just because it's technically possible. What guardrails do you have built into your agent to make sure it doesn't take your Salesforce database and hand it to some random person on the internet? Even if you as the CMO know that — does your VP of Marketing? Does your team? Not everybody needs to be doing the most advanced cutting-edge things possible just because those things exist.

Mandy Hornaday:

So where would you encourage CMOs to start today? Let's say they have a team using the prompting function and they want to move toward more autonomous workflows. What's the framework or practical first step?

Nicole Leffer:

Before you start building, put a few things on paper.

First: look at your workflows. What is the current workflow, and how does it change when you use AI? Where are the places of human bureaucracy that you can strip out? And importantly — where do you want a human to be involved? What are the decision points where you want human judgment?

Second, and this is where I don't see nearly enough attention: what does good look like? Put it on paper. What is good output? What is bad output? You can't start building effectively until you understand that. Once you can articulate what good looks like, then you understand what context the AI needs to create that good output — what information, what guidance from a human. Get those pieces together first. Whether that's in a chat interface or an autonomous agent, those are the things you need, and they're what most people are skipping.

The other thing I'd strongly encourage: if you haven't started building yet, you actually have an advantage. You have the opportunity to build tool-agnostic. A lot of people who started early built their entire structure into a specific ecosystem because they just wanted to get started. The ones with a real advantage are going to have highly portable workflows.

We've seen over the past few years that we don't know who the front runner is going to be. A year ago the following was around one company. A year from now it might be around another one. We don't know what's going to launch by the time we get off this call that upends what we thought we knew. So don't build for any specific tool. Don't be attached to any specific tool.

What happens if the AI tool you've built around triples its price? What if it becomes more expensive than humans? We don't know. So don't put yourself in a situation where you're completely locked into any specific tool.

Right now I'm seeing and hearing about Claude Code everywhere, and Claude Code is an incredible, very powerful tool — I'm not saying anything negative about it. But Codex [⚠️ verify spelling against audio] in the OpenAI system can also do a lot of what Claude Code does. How do I build in a way that I could use it in Claude Code or Codex or something that hasn't launched yet? That way I'm not stuck in any specific ecosystem.

And the practical answer for how you do this is largely about where and how you're storing your documentation. How are you creating AI systems that document their memory in a way that's not specific to one tool? When these tools work on our own computers and in our own files, that actually gets easier.

Anthropic has been really incredible about this — they've open-sourced a lot of their major innovations and made them open standards. MCP, for example — an API connection between an AI and your tools or systems — Anthropic created it but made it an open standard. So now you can use MCP connections with any AI. If you're building your connections as MCPs, that's portable. Similarly with skills — Anthropic made that an open standard too. Now you can use skills across Codex, ChatGPT, and other tools. If you're building in a portable format, you're in a much stronger long-term position.

Mandy Hornaday:

That's so helpful. I hadn't really thought about it that way before. Where do you stand on build versus buy? When does it make sense to use a purpose-built AI tool versus just working within your core LLMs?

Nicole Leffer:

It depends on the company, the team, the specific use case, your skill sets, and your risk tolerance. The vast majority of things you can do with the core tools — OpenAI, Anthropic. Everything else is built on top of that. It's just the wrapper, the harness.

For most things, your core tools are going to be a strong option. There are some specific cases — video, image generation, certain data-related workflows — where going beyond makes sense. But for general-purpose workflow, I don't see the point of buying versus building for everyday stuff.

Here's the reason: when a company is building a purpose-built AI tool, they're building it for lots of different customers at scale. So they're going to pick the cheapest model they can to get to a quality level their customers will tolerate. That's just the economics. Whereas if you're doing it inside your own subscription-based LLM — ChatGPT, Claude, whatever — you can pick the best model for your specific use case. You're optimizing for your one use case, not for thousands of customers at the lowest common cost.

That said, building out full agentic workflows in Claude Code or similar tools can get expensive fast depending on usage. So I'd look at the whole picture and evaluate accordingly.

Mandy Hornaday:

This has been such a great conversation. Nicole, is there anything you'd leave with us? Something you'd tell every CMO to do this week?

Nicole Leffer:

Breathe. Take a deep breath. Take care of yourself in all of this. Take care of your nervous system. This is moving really fast, and it's going to keep moving really fast for a while. Breathe first.

And if you want something more practical: sit down and figure out what you are actually trying to build toward, so you can focus on what's most important to that vision. You cannot build something if you don't know what you are building. Stop. Take the time to ask: what am I ultimately trying to do? And then you can start focusing on the most important steps to make that happen, instead of getting caught up in the noise of what everybody else is doing just because they can.

Mandy Hornaday:

Love it. Thank you so much, Nicole. People can find you on LinkedIn, and it sounds like you also have an online course?

Nicole Leffer:

Yes — I have a Foundations of Generative AI for B2B Marketing course that is extremely popular. You can take it as an individual, and I also have a lot of teams that take it as a group. It gives you the foundational skills to be using tools like ChatGPT effectively. And I do in-house trainings for companies, work with CMOs on their adoption strategy, consulting, advisory — all of that. And I may have more courses coming out soon as well.

Mandy Hornaday:

Can't wait. Awesome — thanks so much, Nicole.

Nicole Leffer:

Thank you so much. Have a great rest of your week.

Mandy Hornaday:

Thanks so much for tuning into this episode of Growth Activated. I hope this conversation sparked new ideas, challenged your thinking, and gave you practical tools to help elevate your impact as a marketing leader. If it did, I would love for you to pass it along to a friend or colleague in B2B marketing. The more we grow together, the more we raise the bar for what marketing leadership can look like. And as always, keep activating growth for yourself and your company. See you next time.