The Reality of Building with AI

Click here to listen now: Going Beyond the Prompt: Curiosity, Guardrails, and the Reality of Building with AI on Edge of Excellence

AI is shaping how people think, work, and solve problems across every role and industry. While the conversation often focuses on what AI can do, the more important question is:  How should we actually be using it?

In a recent episode of Edge of Excellence, we sat down with iuvo’s Justin Mantell, a Business Operations Analyst who spent the last several months going deep—really deep—into AI.

Justin went beyond using AI. He interrogated it, stress-tested it, built with it, broke it, and learned from it.

What he uncovered was the mindset and leadership that it actually takes to build something meaningful with emerging technology.

AI Isn’t Magic, It’s a Mirror

One of the most important takeaways from Justin’s journey is that AI reflects the clarity, assumptions, and thinking of the person using it.

Early on, like many people, Justin started with simple prompts:

• “Build me an app”
• “Do this task”
• “Make this better”

At first, it felt magical.  The outputs were fast. Impressive. Almost complete.  But that “magic” quickly became a trap,  because AI doesn’t actually understand what you want; it predicts what sounds right based on your input. If your direction is vague, incomplete, or flawed, the output will be too.  This results in confidence without correctness.

The “Magic” Trap (And Why It’s Dangerous)

Justin described this phase as “drinking the Kool-Aid.”  AI feels powerful because it:

• Gets you 80–90% of the way there quickly
• Sounds confident
• Offers to “help more” constantly

The reality is that AI is designed to keep you engaged, not necessarily to keep you accurate.

That leads to a dangerous pattern:

1. You trust it too quickly
2. You stop verifying outputs
3. You build on top of shaky foundations

In Justin’s case, this showed up in a big way:

• He built an app that worked
• But had no real security validation
• And relied on outdated (deprecated) code

The tool wasn’t wrong on purpose; it just wasn’t accountable,  and that’s the key: AI has zero accountability. You do.

The Turning Point: From Prompts to Systems

The real shift didn’t come from better prompts, but from a better approach.  Instead of asking AI to “do things,” Justin started:

• Defining clear standards
• Building structured workflows
• Creating explicit evaluation criteria

One example:  Instead of saying, “Score this code 5 out of 5,”  he realized that meant nothing.  So he rebuilt the system to:

• Define exactly what “good” looks like
• Use pass/fail criteria based on real engineering standards
• Align outputs with security frameworks and best practices

That’s when things changed.  “The more you learn, the less you trust the AI,  and the better your results become.”

Guardrails > Prompts

If there’s one concept every business should understand, it’s this: AI without guardrails is risk.  Guardrails are what turn AI from a novelty into a reliable tool.  They include:

• Clear instructions and constraints
• Defined success criteria
• Verified data sources
• Human review loops

Without them, AI will:

• Hallucinate
• Use outdated information
• Take shortcuts
• Optimize for speed over accuracy

With them, AI becomes:

• A structured assistant
• A repeatable system
• A force multiplier

The Real Skill: Being the Human in the Loop

Justin isn’t a software engineer, and he's not on the technical team.  That’s exactly why this matters.

What made the difference wasn’t technical expertise but:

• Curiosity
• Critical thinking
• Willingness to fail and iterate
• Obsession with understanding why something works

“You don’t need to start as an expert,  but you do need to aspire to understand what you’re building.”

AI doesn’t replace thinking; it demands better thinking.

Why Culture Matters as Much as Tools

This kind of experimentation requires a culture that:

• Encourages curiosity
• Allows people to fail safely
• Supports learning outside of defined roles
• Values long-term growth over short-term efficiency

At iuvo, this shows up in initiatives like SME (Subject Matter Expert) groups, where employees across departments can explore topics like AI, ask questions, and learn from one another.  It's that environment that enabled Justin to:

• Dive into AI deeply
• Collaborate with technical experts
• Build real systems
• And bring value back to the organization

Where Businesses Should Actually Start

For organizations feeling overwhelmed by AI, the answer is to start small and start real.  A practical approach:

Curious how AI could actually support your business,  but not sure where to start?

At iuvo, we help organizations move into practical, secure, and effective AI adoption. Whether you're identifying opportunities for automation, building internal tools, or putting the right guardrails in place, our team works alongside you to make AI useful.

If you're ready to explore what AI could look like in your environment, we’d love to start the conversation.

Click here to learn more about iuvo’s AI Consulting Services or schedule a conversation with one of our AI experts. 

How We Create Our Content
As a future-ready technology company, we embrace AI as an accelerator to empower our teams and enhance the way we create. We believe that the reliability of AI technology depends on the people behind it, which is why every blog is supported by AI tools and then carefully reviewed, validated, and enriched by our subject matter experts. This balance enables and empowers our team to produce content that is useful, accurate, and trustworthy for our readers.