Part 1 of the "Pragmatic AI Adoption" series
Not every problem needs AI. The challenge isn't where we can use AI anymore—it's where we should
Over the past couple of years, you may have noticed a recurring pattern in technology discussions. The discussion often starts with:
"How can we use AI here?"
Rather than:
"Should we use AI here?"
At first glance, the difference seems subtle.
But I think it's one of the most important questions organizations need to ask as they continue investing in AI.
Almost every organization today is exploring AI in some form.
Some are experimenting with copilots.
Some are building chatbots.
Others are implementing Retrieval-Augmented Generation (RAG), AI assistants, or autonomous agents.
The challenge isn't the availability of AI anymore.
The challenge is deciding where it actually adds value.
Because not every problem needs AI.
And sometimes introducing AI can create more complexity than it solves.
If a business process can already be solved using:
then AI may not be the right answer.
This sounds obvious, yet many organizations are currently trying to force AI into places where simpler solutions already work.
You may have seen examples where:
Yet AI was added because it felt innovative.
Innovation is important.
But so is simplicity.
When evaluating opportunities, I find it helpful to think about problems in terms of predictability.
Examples:
These are usually best handled through traditional software.
The desired outcome is consistency, not creativity.
Examples:
These may benefit from AI-assisted capabilities, but often don't require full autonomy.
A combination of traditional software and targeted AI can be highly effective.
Examples:
This is where AI tends to shine.
The problem itself contains uncertainty, interpretation, and context.
That's exactly what modern AI systems are designed to handle.
I don't think AI adoption should be viewed as a binary decision.
It's more of a spectrum.
Manual Process
↓
Digital Workflow
↓
Automation
↓
AI-Assisted Workflow
↓
AI Copilot
↓
AI Agent
↓
Autonomous System
One of the biggest mistakes organizations make is assuming they need to move all the way to the right.
In many cases, the optimal solution sits somewhere in the middle.
Sometimes an AI-assisted workflow delivers most of the value without introducing the complexity and risks of full autonomy.
When evaluating AI, most discussions focus on capability.
Few focus on operational cost.
Introducing AI often means introducing:
The question shouldn't simply be:
Can AI do this?
It should also be:
Is AI the most practical way to do this?
Instead of asking:
"Where can we add AI?"
I increasingly ask:
"What is the minimum amount of AI needed to solve this problem effectively?"
Sometimes the answer is a chatbot.
Sometimes it's a retrieval system.
Sometimes it's a workflow with a small AI component.
And sometimes the answer is no AI at all.
I'm excited about AI.
I've spent a lot of time learning, experimenting, and writing about it.
But I also think we're entering a phase where organizations need to move beyond hype and focus on intentional adoption.
Not every solution should become an agent.
Not every application needs a copilot.
Not every workflow needs generative AI.
The organizations that succeed won't necessarily be the ones using the most AI.
They'll be the ones using AI where it genuinely creates value.
In the next part of this series, I'll explore a question many teams are currently facing:
How do you choose between traditional software, RAG, copilots, workflows, and AI agents?
Because choosing the right AI solution may be more important than choosing the right AI model.
AI is becoming increasingly accessible.
That doesn't mean every problem requires it.
The challenge for organizations is no longer whether they can adopt AI.
The challenge is knowing where it belongs—and where it doesn't.
Perhaps the most valuable AI decision we'll make is deciding not to use it when a simpler solution already exists.