Over the past few weeks, investors have grown increasingly concerned about how artificial intelligence (AI) may disrupt various industries — especially software.
Rather than reacting to headlines, I thought it would be helpful to share how we are thinking about this at Paragon Wealth.
Our goal is simple: understand the real risks, identify the real opportunities, and evaluate our holdings accordingly.
Let’s start with a bit of background.
What Are We Actually Talking About When We Say “AI”?
Most of what people call AI today is actually a large language model (LLM).
An LLM is essentially a math-based prediction engine. It analyzes patterns in language and predicts what the next most logical word or sentence should be. Think of it as a super-advanced autocomplete tool.
It doesn’t “think” — it just predicts.
Since ChatGPT’s release in late 2022, these models have evolved quickly. They’ve gone from answering questions to performing multi-step tasks across different software platforms. That advancement has led investors to question whether some traditional software businesses could face disruption.
Instead of speculating, we’ve adopted a structured way to analyze the issue.
The Framework We’re Using
When evaluating how exposed a company might be to AI disruption, we look at four things:
- Pricing Model
- Liability at Stake
- Physical Products
- Data Advantages
Let me briefly explain each.
1. Pricing Model: Per User or Per Usage?
Software companies typically generate revenue in one of two ways:
- Per user (“per seat”)
- Per usage (pay for what you consume)
If AI enables companies to operate with fewer employees, per-seat models could face headwinds. On the other hand, usage-based models could potentially benefit if AI increases system utilization.
We evaluate how a company’s revenue model may evolve as AI adoption grows.
2. Liability at Stake: How Costly Is It If AI Is Wrong?
AI models are not perfect. Even today, they occasionally generate incorrect or misleading information — often referred to as “hallucinations.”
In some industries, a 90% accuracy rate might be manageable. In others, it would be unacceptable.
If a company operates in a high-liability environment — such as cybersecurity, government systems, or mission-critical infrastructure like utilities — accuracy requirements are much higher. That can create a natural barrier to disruption.
Higher-liability environments often have built-in protection.
3. Physical Products: Is There Hardware Involved?
For years, software-only businesses were viewed as superior because they were asset-light and often enjoyed higher margins.
AI may change that dynamic.
Companies that own large-scale physical infrastructure — such as data centers or networking equipment, or companies that sell physical products along with proprietary software— may have advantages that pure software providers do not.
At least for now, AI can’t build physical things.
4. Data Advantages: Is the Data Unique or Public?
Large language models are trained on vast amounts of publicly available data.
If a company’s value proposition is built primarily on public information, AI may replicate parts of that functionality.
However, companies that control proprietary, closed, or highly specialized data sets may have more defensible advantages.
Closed data can create durability.
What This Means for You
Our job is not to predict headlines or react to short-term market sentiment.
It’s to:
- Evaluate businesses rationally
- Monitor changing risks
- Adjust allocations when appropriate
- Maintain diversified portfolios built around durable companies
AI represents a meaningful technological shift. Some businesses may face pressure. Others may benefit. Many will likely adapt over time.
We are reviewing our holdings through this framework and will continue to make portfolio adjustments as warranted by fundamentals — not short-term noise.
If you have questions about a specific holding or how AI may impact a particular industry, I’m always happy to discuss it.
As always, thank you for your trust.
Ricardo