Duane has spent 20 years in education technology marketing, and he’s watched plenty of trends and “this changes everything” moments come and go. Tools like Photoshop, Illustrator, WordPress, Canva have all made professional-level work a little more accessible to people who weren’t specialists. Now, he’s the CMO at a technology company that integrates AI into its product, which means he sees the shift from both sides: as someone who markets an AI tool, and as someone who uses it every day to do his job.
His take on utilizing AI as a productivity tool and as part of your workflow isn’t about hype, or about AI taking your job. It’s about treating a rapidly improving tool the way you’d treat an excited new hire while knowing where that comparison has its limits.
AI Didn’t Change His Workflow Overnight
When AI first showed up in Duane’s workflow, it didn’t transform it, it was just another “magic trick.”
“At first it was just like, ‘Wow, this is pretty cool,” he said. “Like, oh, I can have it almost like a fancy editor for my copy. Like spell check and grammar, maybe say things a little better than I thought I could myself.”
That is most marketers’ entry point. Better proofreading, cleaner sentences. Nothing that changed how we work. But somewhere in the last eight months, that changed for Duane. AI stopped being a nice-to-have and became infrastructure.
It’s no longer kind of this like, ‘Oh hey, that’s cool,’ but it’s something I use every day, all day long. It functions basically like a colleague of mine within the marketing department.
A gradual on-ramp or implementation is a realistic way to introduce AI into your own workflow. You don’t need to change how your team works overnight, just start with the low-stakes stuff — editing, first drafts, summaries — and let the more in-depth and ambitious use cases migrate to an LLM when you’ve built an understanding of its limits and are confident in its knowledge base.
The AI Workflow: From Interview to Web Page in Minutes
Here’s the thing: Duane’s most used AI workflow starts with a conversation, not AI agents running research in the background or a prompt to create a document.
When he wants to turn a subject-matter expert’s knowledge into marketing content, he records the conversation. He lets tools built into Google Meet or Zoom do the heavy lifting for him so he can focus on the important part: the person.
Once the conversation is automatically transcribed, he hands the transcript to AI along with his company’s existing messaging guide and brand/design architecture and asks it for a web page.
I give it to it, and maybe a minute later, I have a web page that’s ready to go, based upon the transcript. I spot check it to make sure everything’s accurate, and then I can put it into HubSpot and push it live.
A conversation becomes a published page within minutes, rather than going through multiple drafts, revisions, and writer’s block. That’s the real possibility of AI at work, not that it replaces the marketer but that it collapses the distance between “someone said something smart” and “that smart thing is live on your website.”
Is AI Really Going to Last?
Duane’s been through enough tech cycles with their flashy new features that burn out quickly to be skeptical of the “this changes everything” claims, so he knows a thing or two when he says this one is real, but there is a caveat.
His perspective: AI is a tool. It looks like magic until you understand exactly how it works, and it will absolutely make some old workflows obsolete, in the same way that the electric lightbulb shrank the candle-making industry. But the more interesting thing is not what it replaces, but what it opens up.
Duane wanted to be a photojournalist in college but couldn’t afford the camera, lenses, film, and other supplies. Digital photography made it much more accessible years later, and he’s seeing AI do the same thing for a much wider range of skills.
You’re going to see ideas and creation, innovation come out of places that people had no business approaching before, because they’ll have this tool that helps them build out an idea and do something amazing. I have no business building web pages in like two minutes that are really awesome — that’s a developer’s job. Yesterday.
That’s the big revelation. AI at work isn’t a story about jobs disappearing. It’s a story about who gets to build things now that couldn’t before. A marketer building a functional web page, a non-designer producing solid layouts, a small team doing what used to require a bigger one.
Stop Treating AI Like Magic. Treat It Like a New Hire.
This is the concept that Duane keeps coming back to, and it’s the most important insight in our conversation with him.
He describes a common pitfall as treating AI like a magician’s magic hat: tap it, and something amazing pops out. Sometimes literally what you asked for. Sometimes a giraffe when you asked for a horse. When it gets it wrong, it’s easy to say that the tool is unreliable.
When you stop treating AI like it’s magic — expecting it to work like magic — and you treat it more like logic; like a very smart, willing, and able colleague or employee, that’s actually when the real magic starts.
His practical version of that looks a lot like onboarding a new employee:
- Onboard it first. Give it context about you and your brand, your mission and culture, before you ask it to produce anything.
- Train it on the specific job. Don’t just say “make something good.” Give it the job description: the objective, the audience, and the tools and resources that it needs.
- Tell it who it works with. Who does it hand off to? Sales, product, another team? What does that relationship look like?
- Keep developing it. Just like a real employee, it needs training: new information, corrections, updated data, etc.
AI isn’t a one-time setup, it’s an ongoing working relationship. The output quality follows pretty closely with how much real context you give it.
The Time It Went Great, and The Time It Didn’t.
Duane’s honest about both sides; at its best, AI can make your workflow smoother and quicker, consistently.
The miss: he once fed a technical efficacy study — full of specific data, terms, and equations from his company’s PhD researchers — into AI and asked for a landing page. It came back and looked great, so he published it without running it past anyone. It was live for a couple of weeks before his CEO called. The page was beautiful, but some of the data was wrong.
I had to reflect and go, ‘No, I gave it something and I acted hastily. I didn’t review it like I would with anyone else I hired… I still have to have a review process. I still need to check it with my experts before it goes live.’ Because it was 90% right.
It wasn’t the output that failed, it was skipping the review process. It’s the same standard he’d apply to a new hire’s first draft on a technical topic. Good work should still be checked by the subject-matter expert before it goes live, especially anything with numbers attached to it.
Key Takeaway: Give AI a Job Description, Not a Wish
Utilizing AI in your workflow isn’t just about learning a new tool. It’s a change in expectation from “surprise me” to “here’s the job, here’s the context, here’s how I’ll check your work.”
Here are a few things to try this week, straight from Duane’s approach:
- Give AI real context. Start with your brand voice, past work, and goals before you ask for an output, the same way you’d brief a new hire.
- Be specific about the job to be done. “Write something good” will get you nothing useful, whereas a clear objective will get you a usable draft.
- Build in a review step. This is especially important for anything technical or data-driven. Speed is a benefit of AI, but accuracy is still your responsibility.
- Keep it updated. Feed it new information as your business changes, the same way you’d keep an employee up to date.
None of this requires an engineering degree, or panicking about AI taking over your job. It requires treating a very capable new colleague like a very capable new colleague. Brief it well, check its work, and let it handle the parts of your job that used to require a specialist you didn’t have.
