The Lego-Building Approach to AI Infrastructure

If you woke up this morning and read about yet another AI model or agent integration and felt overwhelmed, you're not alone.

The AI landscape changes so fast that by the time you've figured out one approach, three new ones have emerged. So how do you build AI infrastructure when everything is constantly shifting?

Instead of trying to build a cathedral overnight with AI, think smaller. Think…Lego.

The Modular Advantage

When you’re building with AI, you’re creating tools that can adapt, evolve, and improve as better technology becomes available. Taking a modular building approach:

🔄 Gives you the flexibility to swap components in and out.

⏫ Lets you upgrade your AI tools without rebuilding your interfaces.

⏩ Empowers you to maintain speed to value while staying adaptable. 

If your AI tools are built from small, discrete “blocks” of code, reorganizing, upgrading, and tweaking them in the future becomes easier. The AI infrastructure you build on day one will inform how easy—or how difficult—it will be to keep up with developing AI tools as they improve. AI requires a “speed to value” approach: how can your technology, tools, and platforms support your company in creating value as fast as possible? A flexible, modular AI tech stack serves speed to value for much longer than a complex AI tool built for a single task without the future in mind. 

Rethinking the AI Build vs. Buy Decision 

With so many software solutions available on the market, it can be hard to determine whether you should buy or build what you need.

If you want  a bespoke solution to a unique problem in your organization, you need AI to deliver maximum speed-to-value through modular, adaptive infrastructure. It starts at the top, with leaders who say, "These are our strategic objectives, and this is how we'll use AI to achieve them." 

Remember Jeff Bezos's API mandate at Amazon back in 2002? He didn't start with the technology. He started with the strategic vision that Amazon needed to share and use data seamlessly to achieve their growth goals. That leadership decision spawned Amazon Web Services and changed how we think about cloud computing.

📣 Your AI infrastructure decisions should be strategy first, technology second.

Finding the Right Infrastructure

 Before investing in AI solution-building, leaders should ask themselves:

❓Could AI infrastructure support our strategic direction?

🔹 Will our AI tools be able to adapt alongside objectives?

🔹 Are we transforming how we work, or just automating existing inefficiencies?

❓How flexible is the AI infrastructure we want to build?

🔹 Can we easily swap, trade, or upgrade components later?

🔹 Are we using the right model for the right task?

❓How will we implement our flexible infrastructure?

🔹How are we measuring the performance and value of our AI infrastructure?

🔹How will our AI tools coordinate to complete complex, cross-departmental tasks?

The Four Pillars of AI Infrastructure

Once you’ve decided to build a solution using AI, it’s essential to understand the four key pillars of AI infrastructure: data, models, execution, and delivery. Think of each pillar as its own set of Lego blocks: if the engine isn’t working, swap out parts until it does. If the data needs improvement, tweak how you approach it. Starting with a modular approach makes it easy to change out elements that aren’t working, and to make important updates as the available technology improves. 

📊 Data. This is your organization's greatest asset, especially if your company is rich in data. Don't wait for "perfect, clean data." Start with what you have and use AI to make it usable. Your unstructured data (customer conversations, employee expertise, behavioral patterns) could be your biggest competitive advantage.

⚙️ Models. Apply the right tool for the job: you'll need different models for different tasks. Models were built for specific tasks, so apply them appropriately on tasks that match their strengths. Don't try to use one model for everything.

🤖 Orchestration. This is where your AI agents do their best work: organizing data, ensuring quality, routing queries to the right tools, and diagnosing problems. Work closely with your models to prime them for the optimal output.

🏆 Consumption. The best AI experiences feel seamless and intuitive: it’s AI that works so naturally they barely notice it's there. Thoughtful user interfaces, self-service, workflow integrations make it so AI is easy to adopt and flexible to adapt to changing user needs.

The goal isn't to have the most sophisticated AI infrastructure. It's to have the most effective one—modular, flexible, and aligned with your strategic objectives.

Because at the end of the day, AI infrastructure isn't about technology. It's about enabling your organization to create value faster and more effectively than ever before.

🗣️ Your Turn

How are you approaching AI infrastructure in your organization? Are you building cathedrals or assembling Legos?

What I Can’t Stop Talking About

• The best approach to fears surrounding AI is to shift the culture of work. The leaders who roll out the tech and forget about people won’t get far. Can empathy, group learning, and celebrating the little wins really make a difference? It turns out addressing fears surrounding AI actually works.
• We already know who the AI leaders of 2030 are. The leaders who can confidently say “I don’t know,” are the ones making strides. 

My Upcoming Appearances

• Aug 29: Indy SHRM Annual Conference Keynote, Indianapolis, IN
• Oct 15: Executive Women's Forum, Keynote, Denver, CO
• Nov 13: Brilliance 2025, Celebrating Women Disrupting Healthcare Keynote, Chicago, IL