Going Beyond the Algorithm with Mike Taylor

Going Beyond the Algorithms with Mike Taylor

In this episode of Smooth Operators Guide by Livedocs, we had the pleasure of chatting with Mike Taylor, Co-founder of Vexpower. Mike has an extensive background in AI, product management, and workflow automation. 

In this edition, Mike and Ani discuss various topics related to Vexpower, generative AI, marketing strategies, mentoring, and emerging trends in AI and product engineering. Mike explains the courses offered by Vexpower and how they provide on-the-job training for technical tasks in marketing and growth. He also shares his journey from running a marketing agency to working on generative AI projects. 

They discuss the impact of generative AI on the marketing landscape, particularly in creative testing and campaign optimization. Mike shares his strategies for scaling Facebook ads and reducing CPA while growing the user base. 

We're thrilled to have him on this edition of our biweekly newsletter to share his expertise. So, without further ado, let's get into it:

Ani: Vexpower looks super interesting, especially with its mix of prompt engineering, marketing, and modeling. As someone with a background in computer science, I often blend technical skills with growth marketing. Can you tell us more about the courses Vexpower offers?

Mike: Vexpower was born from my time at an agency where we had to train people in technical tasks. We noticed there weren’t many good courses for marketers who code or developers focused on growth, so we created our own. The key was making it practical—less about polished slide decks and more about real, on-the-job training.

Each course is designed around a specific problem and solution, something my co-founder James and I have encountered in our careers. It’s like sitting in on a call where I walk you through solving a problem. Vexpower targets technical marketers and developers who enjoy hacking together. 

It’s a lifestyle business for us, so we’ve automated much of it with LLMs, reducing our work to just one day a month. The community we’ve built is passionate about growth and technical problem-solving, and that’s what drives Vexpower.

Shifting Gears: From Marketing to Generative AI

Ani: That sounds pretty cool! I'm really interested in exploring more about your work in AI, especially after this recording. But first, I'm curious—what inspired you to shift from running a marketing agency to focusing on generative AI projects like Bragpool.dev?

Mike: It was definitely a big change. After a career in marketing, I started experimenting with AI in 2020, getting early access to GPT-3. Although it was intriguing, there wasn’t much financial opportunity unless you were a machine learning engineer. I kept dabbling in AI while consulting, but it wasn’t until ChatGPT launched that the demand for AI strategies surged.

At that point, I found myself enjoying AI work more than marketing, even though marketing was lucrative. Last year, I made the bold decision to drop my consulting contracts and focus solely on AI, despite the financial sacrifice. 

It was risky, but with a book deal from O’Reilly and a successful Udemy course, it felt like the right move. Now, things are picking up—the book is out, the course is growing, and it’s all starting to pay off.

Mike’s Book and Udemy Course

Ani: I came across your book, and I’m really tempted to dive in. The Udemy course also sounds intriguing. Could you give us an overview of what to expect from the book and maybe share some practical examples or scenarios?

Mike: When we got the book deal, we knew it would take over a year to produce, which was daunting given how quickly AI evolves. So we focused on what has remained constant in AI over time. We identified five key prompting principles that work across various models, from GPT-3 to GPT-4, and even future models like GPT-5.

These principles form the foundation of the book, providing a comprehensive guide for anyone interested in becoming a prompt engineer. It’s a technical book that walks you through everything you need to know, making it a solid resource for building a strong AI foundation.

The Udemy course, while based on the same principles, is more tactical. It’s updated monthly to reflect the latest trends and techniques, offering a hands-on approach with practical examples. 

The book is more timeless and strategic, while the course is designed to keep you on the cutting edge of AI advancements.

Creative Testing and Campaign Optimization

Ani: I can see myself and my team increasingly relying on generative AI tools like ChatGPT or Claude for brainstorming, creating outlines, and writing copy. 

How do you see generative AI transforming the marketing landscape, particularly in creative testing and campaign optimization?

Mike: Generative AI is a game-changer, especially for creative tasks. I use it extensively to analyze and optimize creatives. For example, I’ll look at competitors’ ads or customer reviews and have AI identify recurring themes or effective hooks. This helps turn unstructured data into actionable insights.

AI also plays a crucial role in market research, making it much easier to gather and analyze data.

I even run scripts that can tag elements in images, helping to determine what works best in ads. With these insights, I generate creative ideas using AI, which I then pass on to a designer or copywriter for final production. 

While the final output isn’t always AI-generated, AI significantly enhances the planning and research stages, leading to more well-rounded and effective campaigns.

Insights from Mike’s Mentoring Experience

Ani: As a mentor with high ratings and having guided over 200 marketers, what common challenges do you see them facing today? How do you guide them through these issues, especially with the rise of AI and the evolving landscape?

Mike: I use Growth Mentor for my mentoring, which started as a way to continue providing advice after leaving Ladder. Many marketers I mentor are often looking for validation and confidence in their ideas. They usually have good instincts but might not be heard or supported by their superiors.

The main challenges include overcoming self-doubt and getting past the hesitation to act.

Many marketers struggle with the fear of not having the perfect idea and end up not doing anything. My role is to validate their strategies, boost their confidence, and encourage action. This reassurance can be a significant factor in their success.

Mentoring also benefits me by keeping me informed about current trends and tactics. It helps me stay updated on what’s working and what’s not, which is valuable when consulting or running campaigns. So, while mentoring helps others, it also keeps me engaged and informed about the evolving marketing landscape.

Ani: Given your expertise across marketing, growth, and AI, what emerging trends in AI and product engineering are you most excited about? How do you see these trends impacting growth, marketing, and product management?

Mike: One of the most intriguing trends is how AI is transforming measurement and evaluation. In both prompt engineering and growth, the challenge often comes down to measuring performance effectively. 

Many organizations struggle with defining what constitutes good performance, whether it’s evaluating a blog post or assessing growth metrics.

Currently, a significant shift is happening where AI, particularly LLMs, are being used to evaluate content and performance. For example, instead of having a human review thousands of AI-generated blog posts, we can use LLMs to assess their quality based on defined criteria. This approach not only saves time but also introduces a level of objectivity and scalability.

Looking ahead, I envision a future where AI will not only generate content but also autonomously evaluate and refine it. Imagine an AI system that can analyze performance data from Google Analytics, establish its own evaluation rules, and continuously improve its outputs. 

This self-improving, agentic workflow will enable AI to act more like a team member—interacting with you via Slack or email, managing tasks, and optimizing performance.

In the next decade, I anticipate that many businesses will employ more AI agents than human employees. These AI "employees" will handle a range of tasks, from content creation to performance evaluation, revolutionizing how we approach growth, marketing, and product management.

Ani: I recently spoke with Jacob Bank, founder of Relay.app, who shared insights on how AI's involvement in content creation should align with the stakes involved. For high-impact content, like major publications, more human involvement is crucial. But for tasks like summarizing meetings, AI is perfectly suited. 

What’s your take on this perspective?

Mike: I agree. AI excels in summarizing routine tasks like meeting notes, where its efficiency and accuracy can surpass human performance. In contrast, high-stakes content requires a human touch to ensure originality and maintain reputation. 

However, as AI models improve, they will increasingly handle more complex tasks effectively. We’re nearing a point where AI will not just assist but will be preferred for many tasks due to its capability to deliver superior results.

Ani: That’s fascinating. I saw a graph showing AI models doubling in performance and halving in cost every six months. How does this rapid advancement impact the industry?

Mike: The rapid advancement is indeed both exciting and daunting. With AI becoming significantly better and cheaper so quickly, it will reshape how we approach many tasks. Businesses will increasingly rely on AI not just for efficiency but for higher-quality outcomes. This trend will likely lead to a shift where AI is preferred for tasks where it excels, leaving humans to focus on areas where their unique skills are most valuable.

Ani: Regarding your recent experiment comparing AI and human performance, what inspired it, and what were the key takeaways?

Mike: The experiment stemmed from my experience with prompt engineering and a desire to assess how AI stacks up against humans. I wanted to see if AI's performance could be accurately judged and preferred. 

The results were surprising—while only about 40% of participants correctly identified AI-generated content, a majority preferred it.

This experiment highlighted AI’s growing competency and the need for people to reassess their roles. AI can handle repetitive tasks effectively, freeing up humans to engage in more strategic and creative work. This shift could enhance job satisfaction and productivity in roles like product management.

Essential Tools and Hidden Gems

Ani: If you had to choose one tool you can’t live without, what would it be? I imagine it might be something essential for building AI apps, like an IDE or a playground.

Mike: Definitely Claude itself. It handles 95% of my work now. However, I also want to highlight Cursor, an IDE based on VS Code. It's like GitHub Copilot on steroids, packed with features that should have been built in by GitHub already. I only adopted it recently and regret not doing so sooner—it's been incredibly helpful in streamlining my tasks.

Ani: I'm also a fan of Cursor! Do you have any undiscovered gems or techniques in your toolkit that deserve more attention?

Mike: I’d suggest a technique rather than a tool. Most people know that including examples in prompts helps AI generate better results. What’s less known is that AI can create these examples itself. For instance, asking AI to generate good examples of social posts, and then using those examples in your prompt, can lead to even better output. 

This approach, often referred to as using synthetic data, allows AI to teach itself and improve its performance. It’s a surprisingly effective method and something I’m now applying extensively.

Final Thoughts

Ani: Any final thoughts or exciting projects you're working on? Where can people find you—LinkedIn, Twitter, or perhaps through a newsletter?

Mike: You can find me on Twitter at @Hammer_MT, which is the best place to reach me. My book, Prompt Engineering for Generative AI, is available on Amazon. It features an armadillo on the cover, which makes it easy to spot. 

I’m involved in various projects and always open to discussing AI, so feel free to tweet me with any questions!

Ani: Amazing. Thank you for joining us on today’s Smooth Operators Guide. Until next time, thank you, Mike!

Epilogue

Mike shares his insights on prompt engineering and the surprising effectiveness of AI-generated examples:

Key Takeaway #1: AI is evolving to autonomously measure and evaluate tasks, potentially becoming integral team members, with humans managing and teaching them.

Key Takeaway #2: AI is transforming SEO, but traditional practices like content quality, authority, and keyword optimization remain essential.

Key Takeaway #3: AI is preferred for tasks like meeting summaries and content generation, allowing humans to focus on more creative work.

Key Takeaway #4: AI tools boost productivity and job satisfaction, while prompt engineering and AI-generated examples can enhance performance.

‍

Thank you for reading/listening! If you found this episode valuable, please consider sharing it on Twitter/LinkedIn and mentioning @livedocs. You can also leave us a review on YouTube to help others discover the podcast.

‍