Back to Blogs
Drew Neisser is the founder of CMO Huddles and a globally recognized authority on B2B marketing. He’s an AdAge columnist, LinkedIn TopVoice, leading CMO coach, podcast host & friend of penguins everywhere.
At The CMO Table in San Mateo, Paige O’Neill of Culture Amp and Micheline Nijmeh of ThoughtSpot shared how AI is reshaping marketing teams from the inside out across strategy and operations. Their lessons were practical, specific, and refreshingly caveated: start early, map the work, govern the agents, protect human judgment, and measure impact carefully.
At The CMO Table in San Mateo on July 29, 2026, more than 100 B2B marketing executives gathered to wrestle with a question that has moved from interesting to urgent: How are CMOs using AI to rebuild the marketing function?
Before diving in, a flocking awesome thank-you to our co-hosts: Pepper, led by Anirudh Singla, and GTM Leader Society, led by Anjai “AJ” Gandhi. They helped pull together a room full of senior marketing leaders who were ready to compare notes, challenge assumptions, and make the AI rebuild feel a little less lonely.
Not “how do we write emails faster?” Not “which shiny AI tool should we pilot next?” And definitely not “how many agents can a marketer build before needing another dashboard, another governance committee, and a lie-down?”
The real question is bigger: How do CMOs redesign the way marketing works without losing the human judgment, customer empathy, and strategic discipline that make marketing worth doing in the first place?
That was the focus of my conversation with Paige O’Neill, CMO of Culture Amp, and Micheline Nijmeh, CMO of ThoughtSpot. Both are deep into the AI rebuild, but neither pretended to have found the magic formula. That may have been the most useful part.
They are testing. Mapping. Training. Governing. Automating. Measuring. Adjusting. Occasionally staring at the bill. And, yes, still keeping humans firmly in the loop.
I opened the panel with a penguin metaphor, because apparently that is now both my brand and my burden.
On land, penguins look clumsy. They waddle. They trip. They seem like evolution briefly stepped away from the keyboard. But in the water, they are astonishingly fast and efficient.
That is where many marketing teams are with AI right now. On land, the rebuild looks clumsy: isolated agents, disconnected pilots, teams experimenting in pockets, governance arriving after the fun starts, and no one entirely sure which workflow connects to which system.
But the promise is that, once marketing gets into the water, the function can move faster, smarter, and with much less drag.
Paige was quick to ground the room in reality. “We’re not in the water yet,” she said. “We’re still on land, so it feels a little clumsy.”
That was the right starting point. The AI rebuild is not a straight line from experiment to efficiency. It is a change-management journey, an operating-model redesign, a governance challenge, and a business-case exercise all at once. Easy peasy, said no CMO ever.
At Culture Amp, Paige’s team started with an unusually granular exercise: mapping the actual work marketing does.
“We took every person on the marketing team, every function on the marketing team,” Paige explained. The team used AI to consolidate signals from Slack, Google Drive, Confluence, documents, and other systems to understand the footprint of marketing work across the business.
From there, AI helped break the work down into roughly 5,000 units. Those units were then organized into 36 marketing scenarios, including workstreams like pricing and packaging, brand asset creation, ABM campaigns, and regional field marketing events.
That level of detail matters because “automate marketing” is not a plan. It is a slogan wearing a headset.
Paige’s team needed to know which tasks were appropriate for AI, which required humans, and where automation would create the most leverage. The result was a clearer view of the work and a hypothesis that about 50% of the marketing function could eventually be automated.
Just as important, the exercise surfaced operational problems that were not strictly AI problems.
Paige said the mapping revealed “areas where there was duplication of effort happening, where teams were working on the same thing, where collaboration was breaking down.”
That is a useful reminder for CMOs: AI does not merely automate the work. It exposes the work. If the process is messy, AI will find the mess. If the workflow is redundant, AI will show the redundancy. If the team is aligned only by heroic Slack archaeology, AI will not politely look away.
Micheline’s team at ThoughtSpot came at the problem from a different starting point. As an AI company, ThoughtSpot had no choice but to lean in early.
“Because we are an AI company to start, just by nature, we didn’t have a choice,” Micheline said. “We had to build agents. We had to have AI as part of our journey.”
That created a different problem: Agent sprawl.
Her team ran a hackathon and built more than 20 agents in four hours. That sounds thrilling until you imagine what happens when every team builds an agent, every agent has a different owner, some agents touch the customer journey, some agents do not connect, and one of the builders eventually leaves the company.
Congratulations. You now have agent debt.
Micheline’s key insight was that the next phase is not about building individual agents. It is about connecting them into workflows.
“Agent building is not the key anymore,” she said. “It was cool six months ago. It was great a year ago. To me, it’s the workflows.”
That is a critical pivot for CMOs. The early AI phase rewards curiosity and experimentation. The next phase rewards architecture. A single agent that writes content may be useful. A connected workflow that notices a bad email, routes the problem, triggers the right fix, updates the right system, and improves the next customer touch is something else entirely.
That is where AI starts to rebuild marketing rather than merely decorate it.
Both CMOs emphasized that AI transformation is not only technical. It is psychological.
At Culture Amp, Paige watched the team move through stages: Fear, suspicion, and then excitement. “I’ve watched them go from being afraid to embrace the technology,” she said, to a place where training, experimentation, and participation in the redesign process created real momentum.
The shift took time. Paige described talking about AI’s potential, bringing in consultants, leading the team through months of training, encouraging experimentation, setting goals, and clarifying where that experimentation would lead.
Her lesson for the room was blunt: “Do not underestimate the journey that the team has to go on, and as leaders, we have to take them on that journey.”
That journey took nearly a year.
This is where some AI strategies get wobbly. Leaders announce transformation, buy tools, ask teams to experiment, and then wonder why adoption is uneven. But if people think AI is a threat, a gimmick, or one more thing piled on top of a shrinking team’s workload, they will not redesign the function with confidence.
Micheline made a related point. ThoughtSpot made a company-level promise that no one would lose their job because of AI. The purpose was to remove fear. But the promise did not mean ignoring role design. When hiring, her team now asks: Can AI do this job first? If not, what should the role become?
That is the new leadership balance: Reduce fear without avoiding the hard redesign questions.
No one in the room needed to be reminded that marketing teams are under pressure. Budgets are tight. Backfills are harder to secure. Growth expectations have not developed a sudden interest in empathy.
Paige noted that Culture Amp’s marketing team had gotten smaller twice over the last year. “Marketing took a disproportionate percentage of the cuts, as marketing often does,” she said.
AI became part of the answer to a hard question: How does the team come out the other side and work more effectively?
At ThoughtSpot, Micheline shared a specific content-team example. The team went from five or six writers to three and still produced significantly more work. “We’ve been able to get triple the output because of AI,” she said.
But she added an important caveat. The team is not lounging around with 10 free minutes and a tiny paper umbrella in its drink. They are still working hard. AI is helping them keep up with the volume of work at a fast-growing company.
That distinction matters. Productivity gains are real, but they do not automatically reduce burnout. In some cases, AI helps smaller teams carry larger loads, which can become its own pressure cooker if leaders are not careful.
Paige also warned that AI cost curves may surprise leaders. Individual ChatGPT or Claude use is one thing. Automating campaign creation, execution, and always-on workflows across the business is another.
“The bill is just now starting to come due,” Paige said. “In the short term, I think the cost of doing those things is going to outweigh the people cost.”
That does not mean CMOs should stop. It means they should model the economics honestly. AI may drive savings, speed, and quality, but consumption costs, tool sprawl, governance needs, and implementation support all belong in the business case.
One of the liveliest themes from the panel was governance. Once teams realize they can build agents quickly, the natural tendency is to build them everywhere.
Micheline said ThoughtSpot hired an AI operations role to manage this emerging complexity, particularly around the customer journey. The point was not to make individual agents work. The point was to ensure that all this AI activity connected across the experience.
Paige described a similar move at Culture Amp. The company formed an AI transformation committee to set governance for agents and prevent “agent proliferation.” Her marketing AI director sits on that committee and helps guide decisions.
That may sound bureaucratic, but it is practical. Without governance, companies risk:
My take: Governance should not arrive with a clipboard and kill experimentation. But it does need to show up before experimentation becomes infrastructure.
Micheline offered one of the clearest practical lessons of the session: “Start early, put together the customer first, and put the journey through the customer’s lens, and then start thinking about automation.”
That is worth taping to the monitor.
Too much AI energy still goes toward acquisition because, well, marketers. But the best AI investments may come from improving the customer journey: Onboarding, education, support, renewal, advocacy, and expansion.
I put a big punctuation mark on Micheline’s point during the session because it deserves one. If more AI investment were directed at improving some part of the customer journey, the outcomes would likely be better. Serve customers better, and prospects tend to come along.
ThoughtSpot also had to solve another AI-created challenge: Product innovation is now faster than marketing’s ability to package it.
“This is the first company, probably as a CMO, where I’m behind the product,” Micheline said. “The product innovation has been crazy because they’re able to innovate at such scale.”
That creates a messaging problem. Customers cannot absorb a new story every week just because the product team can ship one. ThoughtSpot responded by moving toward quarterly launches, packaging capabilities into a story the market can understand.
AI may speed up product innovation, but marketing still has to create coherence. The market does not buy a feature avalanche. It buys a story it can understand.
The audience pushed hard on measurement, as they should.
Omer, who runs B2B SaaS research on benchmarks, asked how CMOs should communicate the impact of AI investments to CEOs and CFOs six or twelve months down the line.
Paige said the topic is already “very top of mind for the board, for the business, for the C-suite,” and that Culture Amp reports AI transformation progress weekly to the executive team.
That is a high bar, and not every organization will be ready for weekly reporting. But the principle is sound: AI transformation needs an executive communication rhythm.
I also asked the room how many could say AI investment had already impacted pipeline. A few hands went up. Then I asked if anyone could show 3x pipeline with the same staff. Crickets, or at least very quiet penguins.
That silence matters.
Many companies are investing money, time, training, and leadership attention into AI. Many are working faster. Fewer can clearly show bottom-line impact yet. That does not mean the investment is wrong. It means CMOs need to be careful about what they claim and disciplined about what they measure.
Paige did share one area with clearer pipeline impact: The SDR organization.
Over the last two years, Culture Amp cut its SDR team by roughly 30%, while SDRs now generate about 90% of pipeline for the business. Paige said SDR productivity improved by about 30% over the last year.
How? A combination of tech stack leverage, AI SDR capabilities, 6sense, Outreach, automated inbound routing, intent scoring, and AI SDRs following up on high-intent leads before passing them to humans.
That example is important because it is not a single magic tool. It is a system. AI created leverage because it was connected to routing, scoring, prioritization, and human handoff.
The audience questions helped move the conversation from inspiration to operating reality.
One audience member asked whether companies need a strong process foundation before they automate. It is hard to automate a process if your messaging, journey, or operating model is still immature.
Micheline’s answer was refreshingly honest: “I didn’t at the time.”
But starting forced the team to identify gaps. “It forced a way for us to think how to do it,” she said. “Our way of building agents and workflows helped us realize there were some gaps that we have to fill.”
Paige agreed that maturity varies by function. “Marketing is never done,” she said. “We’re always a work in process, and I think we can’t wait.” Her key point: If AI reveals that the foundation is weak, do not simply automate the old process. Use the moment to rebuild it.“We’re not just automating the same processes that we had,” Paige said. “We’re looking for new ways to do.”
Another audience member asked about build versus buy, especially as AI systems become more resource-intensive.
Paige described herself as having a “buy first mentality, not a build first mentality as a CMO,” while acknowledging that the balance may change as AI tools and internal capabilities mature.
Her view: There is room for both, depending on the work, partners, and outcomes. Culture Amp will not build everything themselves, and they will continue to use third-party tools where it makes sense.
Micheline, coming from an AI company with its own agent-building solution, has more internal build capacity. Even so, she agreed with the practical spirit of buying where it makes sense: “If you’re going to buy a tool, why work harder [to build your own?]”
The practical lesson: Build when the workflow is strategically differentiated and you have the capability to maintain it. Buy when a specialist can solve the problem faster, better, and with less hidden debt.
One audience member asked whether hallucinations remain a major concern as AI scales across teams.
Paige said she is seeing fewer hallucinations, but the worry has not disappeared. The difference is process. “We’re not just—it’s not going out without human verification,” she said. She also noted that teams are getting better at prompting for verification and building safeguards into workflows.
Her point was practical and calming: Every technology has failure modes. Old marketing automation gave us “Hello, First Name.” AI gives us hallucinations. The answer is not blind trust. It is better validation, human review, and safeguards.
The final audience question asked how CMOs should balance content for human readers and bot readers.
Paige’s answer was encouraging: “The good news is that the LLMs like to read things in the same format that people like to read things in.”
Long-form content, clear structure, and useful formatting help both humans and LLMs. That is a better equation than the old days of writing for Google while trying not to sound like a tax form in a trench coat.
Micheline added a brand warning: “We don’t want us to all sound like AI or bots.” ThoughtSpot is paying attention to how the company shows up, making sure AI-supported content does not become generic.
That may be one of the most important content lessons for CMOs. Structure helps machines understand you. Voice helps humans believe you.
No one has the magic formula yet. Anyone claiming otherwise should be asked to show the spreadsheet, the governance model, the cost curve, and the therapy bill.
But the panel did reveal a practical starting playbook.
Break marketing into actual units, scenarios, and workflows. Look for duplication, bottlenecks, handoffs, and unclear ownership. AI cannot fix what leadership has not yet examined.
Tasks are useful. Agents are interesting. Workflows are where transformation starts. Focus on connected systems that improve meaningful work across the customer journey.
Training, experimentation, psychological safety, and clear expectations matter. Fear does not produce great operating-model redesign.
Create ownership, standards, and review processes before agent sprawl becomes agent debt. Governance should enable scale, not smother learning.
Do not let every AI project chase acquisition. Use AI to improve the end-to-end customer experience, from discovery to renewal.
Efficiency matters, but speed alone is not the business case. Track productivity, cost, quality, cycle time, pipeline impact, conversion, customer experience, and team capacity.
The goal is not to remove humans from marketing. It is to move humans toward strategy, judgment, prioritization, creativity, customer understanding, and decision-making faster.
The AI rebuild is not coming. It is already underway.
But the winners will not be the teams that build the most agents or buy the most tools. They will be the teams that understand the work, redesign the workflows, govern the systems, protect the brand, support the humans, and measure impact with enough honesty to keep credibility in the C-suite.
Paige and Micheline are not pretending this is easy. They are showing what practical leadership looks like in the middle of the mess.
Start early. Map the work. Put the customer journey first. Govern before sprawl becomes debt. Measure impact carefully. Keep humans where judgment matters.
And remember: The CMO role may be the coldest job in the C-suite, but it is warmer in the Huddle.
Start by mapping the actual work your team does. Identify repetitive tasks, duplicated effort, slow handoffs, and workflows that touch the customer journey. Then prioritize automation where it improves speed, quality, or customer experience.
Agents are useful, but workflows matter more. A single agent may solve a narrow task. A connected workflow can improve how marketing operates across campaigns, launches, sales enablement, and customer touch points.
Be clear about the purpose of AI, invest in training, involve the team in experimentation, and explain how roles may evolve. Leaders need to take teams through the emotional journey, not just announce a technical initiative.
Measure efficiency, cycle time, cost, output quality, pipeline influence, conversion improvement, customer experience, and team capacity. Be careful not to overclaim ROI before the data is clear.
One major risk is agent sprawl: Disconnected tools, unclear ownership, inconsistent outputs, and workflows that no one maintains. Governance should begin early, while teams are still experimenting.
CMO Huddles helps B2B marketing leaders win by bringing together peers, fresh perspectives, and opportunities to build stronger personal brands. Want to join the huddle? Learn more about CMO Huddles and apply to join the community.