Every software vendor has added "AI" to their marketing. Every conference has an AI track. Your LinkedIn feed is nothing but AI hot takes. And if you're a business owner trying to figure out what to actually do about all this, it's exhausting.

That exhaustion has a name: AI fatigue. And it's making businesses either rush into bad investments or, worse, ignore AI entirely.

The AI noise problem

The problem isn't that AI doesn't work. It does, in specific, well-defined use cases with good data and clear objectives. The problem is that the hype machine has made it nearly impossible to tell what's real from what's marketing.

"AI-powered" has become the new "cloud-enabled." A label slapped on everything to justify premium pricing. A search box with autocomplete is now "AI-powered search." A rules-based chatbot is now an "AI assistant." A dashboard with trend lines is now "AI analytics."

What AI fatigue looks like

In the businesses I work with, AI fatigue shows up as:

  • Buying AI tools that nobody ends up using
  • Starting AI "proofs of concept" that never reach production
  • Assuming AI can't help because the last AI thing they tried was disappointing
  • Decision paralysis, waiting for the "right" AI tool instead of just solving the problem
  • Vendor fatigue, getting pitched the same AI story from every supplier

Where to actually invest

Ignore the categories and look at the problems. AI investments pay off when they do one of these things:

  1. Automate high-volume repetitive tasks: Document processing, data extraction, form handling, invoice matching. These have clear inputs, clear outputs, and measurable time savings.
  2. Surface information faster: Internal search over your own data, customer self-service over your knowledge base, instant answers from your policy documents. Saves hours of searching and asking around.
  3. Handle the predictable part of customer interactions: FAQ responses, booking management, order status queries. Frees up your team for the complex conversations that actually need a human.

Notice what these have in common: they're boring. They solve real operational problems. They have measurable ROI. That's the pattern to look for.

Where not to invest (yet)

  • "Strategic AI" that's really just a chatbot with a nicer interface
  • AI tools that require your team to change how they work entirely because adoption failure is the number one killer of AI ROI
  • Any AI project without a clear success metric because "we'll know it when we see it" is not a metric
  • Generative AI for customer-facing content without human review because the reputational risk isn't worth the time savings

A simple investment filter

Before committing to any AI investment, run it through these questions:

  1. What specific problem does this solve?
  2. How are we solving it today, and what does that cost?
  3. What's the measurable improvement we expect?
  4. Do we have the data it needs?
  5. Who will own it, monitor it, and fix it when it breaks?

If you can't answer all five, you're not ready to invest. And that's fine. Saving money by not doing the wrong AI project is its own form of ROI.

For a structured framework, see our guide on estimating ROI from an AI assistant and the questions to ask before starting an AI project.

Kasun Wijayamanna Founder & Lead Developer Postgraduate Researcher (AI & RAG), Curtin University - Western Australia
AI StrategyAI ROIBusiness OwnersIT Leaders