Remember Malcolm Gladwell’s rule that mastery requires 10,000 hours of deliberate practice? Hold that thought.
Most of us outside of AI development got our first taste of natural language AI when ChatGPT launched in late 2022. Do the math: fewer than 7,000 working hours have passed since then, leaving even the most obsessive AI-dabbler still 3,000 hours short of expert status.
Someone forgot to tell LinkedIn. Scroll for five minutes and find that your entire network of marketing and GTM leaders has achieved full AI mastery! We’re building apps, deploying agents, and transforming every discipline imaginable. How do you like them apples, Malcolm?
Maybe that’s an exaggeration, but I’m betting there are a lot of marketers out there feeling massive pressure to keep up, and maybe keep a tad quieter on those team calls discussing the latest AI tips and tricks.
If you’re job-hunting, it’s even worse. You know recruiters and hiring managers have an AI BINGO card in front of them and will only consider moving you on in the process if you’re name-dropping the trendiest tools.
And heaven help you if you’re job hunting over 40 and mention that you use ChatGPT. “C’mon, Boomer, it’s all about Claude Code or {insert trendy AI no-code platform here} nowadays – keep up!”
AI imperfection isn’t the same as uselessness
Sure, AI still makes plenty of mistakes or false assumptions when working for you. When providing advice or recommendations, it may give equal weight to a random blogger’s opinion as it does to a thoroughly researched scientific paper if it helps to answer your prompt. Stories of hallucinations and outright errors are shared by our networks daily. It recently even made a simple math error for me when I was having it build an agent to automate a client proposal process, where it calculated 8 x 250 = 2,400. (Spoiler alert: that’s incorrect)
But these mistakes do not mean you should avoid using AI. It simply reinforces the message that you, the human being, are responsible for checking its work and ensuring that the resulting recommendation, process, or application is accurate and usable. If you turn in shoddy work to your leadership because AI made a mistake, it’ll still be you who will be rightly blamed.
Even in these disruptive times, I believe that common sense and work ethic will prevail over sloppiness for most marketers, and overall, AI is a powerful and exciting innovation that has already disrupted everything we know about marketing.
Take a deep breath and dive in
I fully expect the vast majority of B2B marketers already use and find value in AI as a writing or research assistant. Maybe it’s fine-tuning an email to a colleague, drafting a blog post, or helping to compile competitive and market insights. Where the intimidation begins is when people talk about creating “AI autonomous agents executing multi-step workflows.”
But relax, while it may be an accurate technical description of what you’re building, it’s also just a bunch of tech jargon. I asked Claude to simplify what that really means, and it came up with a perfectly good analogy:
Normally, AI is like a vending machine. You press a button, and it gives you one thing back. You ask, it answers. Done.
An “autonomous agent running a multi-step workflow” is more like hiring an intern. You give them a goal (“find me the top 10 competitors and summarize their pricing”), and they go figure out how to do it. They search, read, compare, organize, and then come back with finished work. You didn’t tell them every step. They decided the steps, did them in order, and handled problems along the way.
I really like the intern analogy, because like a human intern who has never performed this task before, the AI may not actually know the right things to search for or the things from their research that you will find most valuable. Just as you would provide guidance and coaching for your intern, you must also validate and iterate your instructions to the AI to fine-tune these processes.
Different AI interfaces for different levels of experience
But like everyone else, I started my AI journey with the simple conversational chat interfaces. And I’m still able to do quite a bit of my work just with that Q&A experience. But when I wanted to build repeatable processes because I already knew all the steps I wanted to execute, I started experimenting with tools like Claude Cowork, Notebook LM, and Perplexity, to name just a few.
From a coding perspective, I personally have no computer science or engineering experience. But I am a data guy, so I do know my way around a SQL query and have been in the B2B tech space, working alongside product managers and developers long enough that I’m not afraid to ask questions. So when I first started using Claude Code, it was more of a leap of faith where I accepted all of its requested permissions without caring what I was permitting. I’m using it for my personal business, so my risk is low. If you’re working for a company, be sure to work with your IT and security teams to better understand what’s allowed and what’s a no-no. At the moment, there is still not a single tool that supports all your use cases just as effectively, so do your research, pick, and choose the right tool for the right project.
Here’s what I’ve built in just the past 2 months:
- GTM “Right to Win” Play Prioritization application. As I shared here, I trained Claude Code on my methodology, process, and scoring model that I spent many years developing. It is now a standalone HTML application that can be installed at any of my client sites, where it manages the intake and scoring process in a much more user-friendly approach than my old Excel spreadsheet.
- Client Proposal Generation Agent. I fed my client proposal template and past examples of completed proposals that I created into Cowork, and now have an intake form where I capture all the key aspects of a potential engagement. When I feed that intake form into the agent, it automatically generates a branded, comprehensive, and ready-to-ship proposal/statement of work. Saves me hours manually writing these proposals.
- First Analyst Briefing Deck Generator. Using Claude Code, I’ve built an application that allows product marketers to share any number of corporate positioning and messaging docs, websites, roadmaps, and any other relevant information, and generate a draft PowerPoint deck and a comprehensive Word document outlining the key messages they should share with the industry analyst with areas that are risks or missing – all based on my training the engine with my briefing template, experience, and writing on this topic.
- Product Launch Calculator and GTM planning agent. I’ve once again trained Claude on the best practices I’ve developed over the years on how to effectively plan, tier, and operationalize a product launch and campaign, and build a repeatable tool to support intake and recommendations for launch teams.
And many more are in development. It’s fun and gratifying to build these tools. I’m spending plenty of time manually ensuring they are working as designed, but any initial fears of feeling ill-equipped quickly faded as I experienced how fast I was able to learn and deliver.
Note: Contact me if you’d like to try any of these tools for yourself. I’d be happy to share and get your feedback!
What to expect moving forward
As you can see from some of the tools I’ve personally developed above, there’s good news for the more experienced, tenured marketers out there (Yes, especially the over-40 crowd). The more personal experience you have in executing these tasks, creating these deliverables, managing complex processes, and building compelling messaging or campaigns, the more valuable and substantive your inputs to the AI tool – and the more valuable and differentiated your results.
The founders and marketing leaders who believe they no longer need experience in marketing because AI can replace that experience will quickly find themselves in a sea of sameness. After all, the inexperienced AI jockeys will be building the same tools as everyone else – all sourced from the single opinion of that random, unnamed blogger!