Hiring a Team of Generative AI Experts

In 2010, I was involved in one of the pioneering successes in the realm of Natural Language Generation (NLG) backed by venture capital, specifically with Automated Insights. I developed a key component of our platform and oversaw the area that harnessed AI expertise beyond just coding.

Despite being a developer, my role was to recruit a skilled team capable of transforming data into meaningful narratives—insights that transcended mere data analysis. My goal was to identify individuals who could not only leverage AI effectively but also excel in it.

During those early days, when NLG was just emerging, I had no clear idea of the skills that would be necessary for success. It was a journey filled with experimentation and learning.

The Significance of These Skills

I strongly advocate for no-code and low-code platforms. A point I often raise—which sometimes sparks debate with traditional programmers—is that everyone, including myself, is essentially manipulating existing lower-level code.

In 2023, the majority of us are not writing low-level machine language; we are advanced users.

So, can nearly anyone learn to code? Perhaps, but it depends on the depth of understanding one desires.

What distinguishes exceptional coders from mediocre ones are specific skills unrelated to rote syntax memorization. The qualities that make code elegant and reusable stem from soft skills that lead to innovative, efficient solutions deserving of great coding.

In recent years, no-code and low-code frameworks have evolved into visual alternatives to textual coding, rapidly closing the gap with traditional syntax.

Will no-code ever fully supplant real code? That’s not the point. Let’s instead delve into the foundational skills necessary for excelling in no-code, traditional coding, or AI.

The Skills Required Beyond Coding

Let’s acknowledge the obvious for those who are purists: coding is still essential to implement the solutions derived from these skills, whether you pursue lower-level coding or higher-level no-code solutions.

If exceptional code stems from great solutions, then great solutions originate from effective problem-solving. However, I won’t oversimplify this. I tackle numerous issues daily, and only a small fraction pertain to coding.

Here are some vital problem-solving skills that contribute to developing the outstanding solutions that culminate in effective AI:

Understanding the Data-Driven Problem

This may seem straightforward, yet several questions can often be answered without grasping the underlying reasons:

• What specific problem are you addressing and what adjacent issues exist?
• Who else is attempting to tackle these challenges, and how?
• To what extent does your data resolve these problems, and how effectively?
• How does your organization derive profit from this solution?

Automated Insights primarily served the sports sector at first. My initial lesson was that individuals who comprehended how data resolved sports-related challenges were far more likely to succeed than either seasoned data scientists or non-technical sports professionals.

Application of Business, Product, and Customer Theories

What happens when you gather a group of machine learning experts, data scientists, and technologists in one room?

This isn’t a joke; we actually did it. The outcome was surprisingly unproductive.

Once we gained market traction, I assembled an innovation team to brainstorm the future of AI. This led to many exciting and innovative ideas, but none were grounded in business realities, nor could they be effectively realized as minimum viable products.

While scientific exploration is fascinating, what is created in a lab often remains there without significant effort to implement it in the market. We relied on individuals who could develop exceptional products, quickly bring them to market, and prioritize customer success.

Focused Creativity and Idea Implementation

Everyone believes they are creative and has brilliant ideas—and they are correct. However, this belief is meaningless unless creativity is directed, and ideas are effectively executed. Typically, the only way to enhance these skills is through trial and error.

At Automated Insights, we tested several product iterations before discovering the application of our idea that resonated. Currently, I am on the third iteration of a recent project after the first two attempts—rooted in the same creative foundation—failed due to poor execution.

Recognizing Patterns and Spotting Trends

If I were to identify a key strength in my coding or AI capabilities, it would be this.

We are currently in a phase where Generative AI is widely accessible, often producing “ChatGPT”-style outputs merely for data regurgitation.

However, true expertise lies in identifying significant trends, changes, and milestones within that data—something I strictly prohibited at Automated Insights. The challenging aspect of our work was discerning meaningful patterns and milestones from data. While any aspiring data scientist might write formulas to mimic this behavior, the ability to visualize patterns and recognize trends is a crucial soft skill for generating valuable AI output.

The Next Steps: Developing State Machines

This skill is the most aligned with traditional development.

The most effective AI applications not only recount past events but also provide recommendations for future actions. In conventional coding, this is managed through state machines, illustrated as flowchart-like diagrams that outline the conditional logic of computer operations.

This concept was foundational to our NLG platform, resulting in a complex network of state machines that facilitated the generation of insightful and valuable content.

If you can visualize, document, and create state machines, you are on the right path to understanding how to navigate the multitude of intricate and interconnected decisions our minds continuously make.

And that is the essence of AI, operating beneath the surface of the code.

This article was initially published behind Inc. Magazine's paywall, where I contribute a column on startup advice and culture. For free access to my public writings, consider joining my email list at joeprocopio.com.

Teaching Startup is my initiative aimed at transforming how startup and business leaders receive advice. It offers hundreds of vetted expert responses to genuine startup inquiries—no fluff, no scheduling, no small talk. Sign up for a free trial of our newsletter.