Beyond Content Creation: Expanding the Capabilities of LLMs in Construction Industries

One of the biggest hurdles for construction organizations looking to harness the power of LLMs is the fact that many people just don’t know what to do with them.

It’s not that they don’t know how to use them—using them couldn’t be easier—it’s that they don’t know what to use them for.

In a sense, it’s the simplicity of LLM’s like ChatGPT and Claude that make them difficult to understand—at least for new users just beginning to experiment.

At this point, large foundational models have been around long enough that nearly everyone knows they can help with content creation, but beyond that, many people are left scratching their heads—or worse, abandoning the technology altogether after asking the LLM to write a social media post, or an email, and watching it produce a terrible one.

Many people don’t stick around long enough to understand why the LLM produced what it produced, or to learn that content creation is perhaps the least interesting thing that LLMs can do.

Perhaps nowhere has this simplicity paradox been captured more perfectly than in the recently released Chapter Two of the book-in-progress, Our AI Journey, published in partnership with Harvard Business Review.

“And what’s more,” the authors write, “the form the technology currently takes is hauntingly simple. It’s just a blank input chat box, with a blinking cursor, staring at you, asking ‘how can I help you today?’…The temptation is to use it like a similarly simple user interface design, a search box from Google or Bing, for example. But that’s like hiring a research assistant, a ghostwriter or consultant, and asking them to tell you what the weather is like outside. Sure, they can do it, but it doesn’t begin to leverage their true potential.”

Later, they arrive at a sentiment that I’ve heard quite a few times over the last year.

“It was ironic. The most powerful software ever developed by humans, and it didn’t even come with a user manual.”

I suppose a user manual would have been nice (if you really want one, you can ask the LLM to create one for you), but my guess is that most people wouldn’t have read it anyway.

They still would have clicked around, and asked their friends, and Googled, and ultimately, ended up in the same place they are now: staring at the LLM’s blinking curser, not sure of what to do.

One of the keys in understanding the power of LLMs is experimenting.

Lots of LLM evangelists (me included) can tell you what they can do with them, but until you develop your own specific use case, until you experience the transformative power for yourself, it’s all going to present as noise.

Wharton Professor and AI expert Ethan Mollick likes to say that “you haven’t really experienced AI until you have had three sleepless nights of existential anxiety, after which you can start to be productive again.”

I tend to agree with him on that point.

When you begin to experience the true power of the technology, and when you start to think about the LLM as the most talented, educated, energetic employee you’ve ever hired, who’s read everything and knows everything and can do everything, you begin to grasp the enormity of what it would mean if everyone in your organization was using LLMs to their true potential.

If you had access to a team of McKinsey or Boston Consulting Group advisors, what would you ask them to do?

What if you had access to a world-renowned ethnographic researcher?

How about a Stanford-educated marketing consultant with a decade of experience advising construction companies on how to grow revenue without adding to their headcount?

My guess is you would ask them questions. You would ask them many questions. Specific questions too, about your unique business, about your unique challenges, about your unique market.

Eventually, you’d ask them to deliver something to you that you could use: a strategy deck, an ideal customer profile, a market analysis, a marketing plan, a new-customer acquisition strategy, or even some suggestions for how to optimize your homepage.

As an example, I took a screenshot of John Deere’s homepage, pasted it in the context window of ChatGPT-4, and asked for five ways to improve the page.

Here’s what it gave me.

Based on the John Deere homepage screenshot provided, here are five suggestions for enhancement:

1. Value Proposition Clarity: Ensure the main value proposition stands out more prominently. If the current headline and sub-headline don’t clearly state the unique benefits John Deere offers, they may be rephrased for better impact.
2. Visual Hierarchy: Review and possibly enhance the visual hierarchy. The call-to-action button could be made more noticeable through size, color, or placement adjustments.
3. Product Imagery: While the product images are clear, they could benefit from being displayed within real-world usage contexts to help users visualize the products in action.
4. Navigation Clarity: Consider evaluating if the navigation bar can be simplified or made more intuitive, ensuring that new visitors can understand the site structure at a glance.
5. User Engagement: Incorporate interactive elements, such as a chatbot for immediate assistance or engaging, dynamic content that highlights key products or services.

These are pretty general suggestions, and there’s nothing mind-blowing about these. But it’s a great start and gives me a good jumping off point.

I was intrigued by the first one, so I then asked it to give me some specific ways to create a clearer value proposition and to give me three examples of value prop statements.

Here’s what it suggested.

To create a clearer value proposition, the statement should succinctly summarize the benefit of your offer to the customer, how you solve their needs, and what distinguishes you from the competition.

Here are three examples of value proposition statements:

1. Experience the Power of Efficiency: John Deere’s advanced technology delivers unmatched fuel economy and durability, ensuring you’re ready for tomorrow’s challenges.
2. Innovation in Every Field: Discover how John Deere’s smart farming solutions can increase your productivity and profitability, sustainably.
3. Dependability Meets Performance: Choose John Deere for equipment that works as hard as you do, backed by industry-leading support and service.

Each of these statements is focused on the core benefits to the customer, emphasizing attributes such as efficiency, innovation, and reliability that are likely to resonate with John Deere’s target audience.

These might be right, and they might be wrong—I don’t work at John Deere, so it’s not for me to say—but the point here is that I started with a screenshot of John Deere’s homepage, and within five minutes, I was working on a new value proposition, which is high-level strategic work that will ultimately lead to something else.

Now imagine if we dug into the CTA strategy, or the information architecture, or the visual hierarchy.

You can see just how quickly LLMs can help hone in on high-value initiatives.

The way to extract the most amount of value from LLMs is to use them for literally everything you do—and to experiment.

They will be good and some things and not so good at others, but you’ll begin to develop a working relationship with the technology that allows you to level up nearly every area of your work life.

By approaching LLMs with a spirit of experimentation and a willingness to integrate them into your workflows, you can unlock a wealth of insights and solutions that were previously out of reach.

So stop thinking about what you can write.

And start thinking about what you can do.