Posts

Scaling AI: How CMOs Master the 4 Shifts

Liza Adams · October 10, 2025 ·

There’s a moment in AI adoption where the question changes. You stop asking “should we try this?” and start asking “how do we scale what’s working?”

This week, I led a half-day workshop at Hellman & Friedman’s CMO Summit in NYC for CMOs in its portfolio companies like Baker Tilly, Zendesk, Athenahealth, Allfunds, Circana, Claritev, Enverus, PointClickCare, Checkmarx, and more.

We focused on driving change confidently with grace and compassion—people-first, AI-forward. We worked through four shifts critical in scaling AI:

From answer engine to thought partner

Go beyond simply asking AI to give an answer. Start using it to pressure-test your thinking and find blind spots.

From faster to better to different

Speed is table stakes. Quality separates you from the noise. But the real opportunity is doing work that wasn’t possible before.

From tools to teammates to workflows

Individual AI use is OK. But you get more value when you orchestrate AI teammates into full workflows that handle entire processes.

From org charts to work charts

Map how work actually flows across teams (breaking down silos), then design AI teammates to fit into those workflows with humans guiding them.

Lots of great discussion about changing buyer behaviors, real-life human + AI org transformation case studies with results, and practical learning seeing AI in action.

Then we traded slides and laptops for chef’s hats, aprons, and spatulas. Turns out execs transforming marketing orgs can also nail a perfectly seared salmon and Brussels sprouts with tamarind glaze, especially when there’s competition and 80s music.

And because Misty was determined to win, she worked with ChatGPT to write us a song about our cooking. Set to “Under the Sea” from The Little Mermaid, we performed it. The video will never see daylight. 🤪

A big thank thank you to Heidi Melin, Esmeralda Pelayo, and to the entire H&F team for your people-first values and for this annual CMO event. Heidi, you are truly an inspiration to CMOs.

Renée Gapen, Sarah Gavin, John Flannery, Karolus Viitala, Keri Brooke, Megan Hanley, Sabine Mullin, Misty Muscatel Davis, Scott Gainey, Piper Turner, Thomas Krauße, Stacy Simpson, Katherine Sloan, Melissa Humphrey, loved spending the last couple of days with you. You reminded me why I love this work. Looking forward to continuing on this AI learning journey with you.

And Karolus from Finland is right: this is hands down the most amazing, awesomest, best in the world CMO Summit. IYKYK.

P.S. Want to see where you are on the 4 shifts? I built a quick interactive assessment that gives you instant guidance on what to do next. Link in comments.

See original post here

3 Ways to Drive Real AI Business Impact

Liza Adams · October 9, 2025 ·

The gap between AI experiments and real business impact is where most teams get stuck. They’ve got scattered pilots and individual wins. What they don’t have is a strategy to drive adoption across the team or prove ROI to leadership.

The teams that close this gap do one, two, or all of these:

► Anchor to strategic initiatives

Map your AI use cases to initiatives that already have owners, budgets, KPIs, timelines, and executive visibility. This increases your probability of adoption and success because you’re working with built-in momentum and resources.

One CMO had “Accelerate Customer Growth” as a top priority with a goal to increase net revenue retention by X percent.

The team mapped AI use cases directly to it: spotting upsell opportunities through behavior analysis, personalizing nurture campaigns, and flagging at-risk accounts early. Because these supported a board-level initiative with clear metrics, they got immediate resources and fast adoption.

Sometimes one initiative needs multiple AI use cases working together. Other times, a single use case moves the needle on several initiatives. Either way, you turn AI from a side experiment into a strategic asset.

► Solve biggest pain fast

Focus on work that’s eating up hours, blocking progress, or burning budget. The bigger the pain, the faster you’ll see value.

A global marketing team I worked with was spending tens of thousands of dollars monthly on agency fees to translate and localize every customer-facing asset into eight languages. The process took weeks.

A small group of field marketers who were local language speakers built custom GPTs with brand guidelines, market expertise, and localized examples. Within a week, they had a working solution. AI handled 80-90 percent of the work, with the team checking it and handling the final edits. Translation time dropped from weeks to days, saving thousands in agency fees every month.

Big pain, fast ROI, and proof that’s hard to ignore. The win inspired the team to quickly tackle other big problems.

► Let your trailblazers inspire what’s possible

Your AI trailblazers have been experimenting, building workflows, and learning what AI can and can’t do. They understand AI’s potential and limitations. They’re pushing it to its full capacity.

Find them. Give them space to prove what’s possible. Then let them mentor the team and show others what they’ve built.

At one B2B tech company I guided, a small group of trailblazers quickly grew to 75 across marketing who became the engine for transformation. They created 211 custom AI teammates during experimentation. Today, 57 are integrated into regular workflows across the team. Their wins moved the skeptics faster than any exec mandate could.

The bottom line: Tie AI use cases to strategic initiatives with momentum. Prioritize high-pain, high-value problems. Let your trailblazers inspire what’s possible.

That’s how you close the gap between experiments and real business impact.

See original post here

Scale Your Expertise: Make GPTs User-Friendly

Liza Adams · October 8, 2025 ·

You share your custom GPT with the team. They open it, stare at the blank prompt box, and freeze. Nobody knows what to ask.

Custom GPTs are AI teammates you build and train with your own knowledge, frameworks, and instructions to do a specific task.

The problem is that most people build custom GPTs the way they’d write instructions for themselves. But you already know what questions to ask. Others using it don’t.

If you want to extend your impact, the GPT needs to work for people who don’t have your expertise.

Below are five ways (when you build them) to make your custom GPTs more user friendly.

When you design for people who don’t think like you do, you multiply your impact. Your knowledge reaches more people. Your work gets used more consistently.

Building a powerful GPT is one thing. Making it easy to use is what actually scales your expertise.

See original post here

Unexpected: Claude 4.5 Swears, Defends Itself

Liza Adams · October 7, 2025 ·

Anthropic’s Claude Sonnet 4.5 just swore and defended itself. 🤣 All good and it’s a decent defense. But it’s so uncharacteristic that it surprised me and gave me a chuckle.

See original post here

Name Your AI Teammates Right: Get Them Used!

Liza Adams · October 7, 2025 ·

Too many AI teammates get built and barely used. The problem is the name, not the tech.

A “Sales Assistant” that’s really just a lead scorer. A “Helper” that isn’t clear what it helps with. An “Assistant” that tries too hard to sound human.

I’ve been working with go-to-market teams who are building AI teammates. These are custom AI tools you build, train, and manage for specific tasks or workflows using Custom GPTs, Claude Projects, Gemini Gems, or Copilot Agents.

At first, I thought naming them would be the fun part. Star Wars characters. Marvel superheroes. The Office references. Turns out, fun names can work but only when they match the relationship you’re creating.

Here’s what I’m seeing.

1) Tool

Name it after what it does, like Lead Scorer, Data Categorizer, and Testimonial Finder.

Pure function, no personality. People use it once, get what they need, move on. Clear and to the point.

2) Sidekick

Name it to signal collaboration, like Draft Helper, Campaign Partner, and Strategy Assistant.

It adapts and works with you without pretending to be someone specific. This is where fun names can work like “Robin” or “Chewy” tell you it’s a helpful collaborator.

3) Persona

Name it after a specific person or role, like LizaGPT (my digital twin), CEO Jordan, Enterprise Buyer Persona, and Industry Analyst GPT.

Less about tasks, more about extending someone’s thinking. It tests ideas, challenges logic, finds blind spots. The name tells you whose perspective you’re getting. “Yoda” works here because everyone knows he challenges your thinking.

Match the name to the relationship you’re creating.

Better names set clearer expectations. And clearer expectations mean your AI teammates actually get used.

Where do your AI teammates sit on this spectrum?

See original post here