There’s a phrase that shows up in almost every AI conversation:
“Best-in-class.”
It’s used to describe platforms, models or vendors often as the primary justification for investment. And on the surface, it makes sense. If you’re going to invest in AI, why wouldn’t you choose the best?
But here’s what most organisations discover too late:
“Best-in-class” AI on top of “worst-in-class” foundations is one of the most expensive mistakes you can make.
Not because the technology is flawed, but because it’s being asked to operate in an environment it was never set up to succeed in.
The Illusion of Capability
Modern AI platforms are incredibly capable.
They can interpret intent, generate responses, orchestrate workflows, and integrate across systems. In a controlled demo environment, they perform exceptionally well.
But production environments are not controlled.
They are messy, inconsistent, and full of edge cases. And this is where the gap between capability and performance becomes obvious.
The issue isn’t what the AI can do.
It’s what the environment allows it to do.
Where It Starts to Break Down
When AI underperforms, the symptoms are usually easy to spot.
Customers receive inconsistent or incorrect answers. Journeys feel disjointed. Escalations happen too early, or too late. Agents are left to recover conversations that have already gone off track.
At first glance, it looks like a technology problem.
In reality, it’s almost always a foundation problem.
Three issues come up again and again.
The first is knowledge.
In many contact centres, knowledge is fragmented across systems, duplicated in different formats, and inconsistently maintained. Even when the right answer exists, it may not be structured in a way the AI can reliably use for the purpose it is meant to.
The second is process.
Where processes are unclear, undocumented, or reliant on agent judgement, AI struggles. It needs defined pathways, clear inputs, decisions, and outcomes. Without that, journeys become unpredictable.
The third is integration.
AI relies on access. Access to systems, to data, to context. When integrations are shallow or incomplete, the AI is forced to operate with limited visibility. That’s when experiences start to break down.
None of these are problems a better model will fix.
AI Performance Is Constrained by Its Environment
A useful way to think about AI is this:
It doesn’t operate at the level of its potential. It operates at the level of its environment.
That environment is defined by:
- The quality of your knowledge
- The clarity of your processes
- The structure of your data
- The strength of your integrations
If any of these are weak, they become the limiting factor. Regardless of how advanced the technology is.
This is why two organisations can deploy the same platform and see completely different results.
One sees value quickly. The other struggles to get basic performance.
The difference isn’t the AI.
It’s everything around it.
The Cost of Getting This Wrong
The impact of weak foundations isn’t always immediate, but it does compound quickly.
You see it in performance first. Lower containment. Higher escalation. Poorer customer experience.
Then, you see it in behaviour.
Agents lose trust in the AI and start working around it. Customers learn to bypass it. Internal teams spend more time managing issues than improving outcomes.
Eventually, the narrative shifts.
What started as a strategic investment becomes something that’s “not quite delivering.” Confidence drops. Momentum slows. In some cases, the platform gets replaced without ever addressing the real issue.
This is where the true cost sits.
Not just in the technology itself, but in the lost opportunity to get value from it.
Why “Best-in-Class” Becomes a Distraction
Focusing too heavily on selecting the “best” platform can distract from the work that actually matters.
Because once the technology is chosen, there’s an assumption that performance will follow.
But technology selection is only one part of the equation. And what might be a controversial view, it is not the most important one.
A well-designed environment with average technology will outperform a poorly designed environment with the best technology.
That’s not a limitation of AI.
It’s a reflection of how dependent it is on the system it operates within.
What High-Performing Organisations Do Differently
Organisations that get value from AI tend to approach it differently.
They don’t start with the platform. They start with the environment.
They invest time in:
- Cleaning and structuring knowledge so it can be reliably used
- Defining processes clearly, end-to-end
- Understanding contact drivers at a detailed level
- Designing integrations as part of the solution, not as an afterthought
They treat AI as part of a broader system, not as a standalone capability.
And because of that, it has been set up to succeed. So, when the technology is deployed, it performs as expected.
A More Useful Question
Instead of asking:
“Which AI platform is best?”
A more useful question is:
“Is our environment ready to support AI?”
Because until that answer is yes, the platform you choose will have far less impact than you expect.
The Practitioner Reality
AI doesn’t fail because it isn’t powerful enough.
It fails because it’s deployed into environments that aren’t ready for it.
Fragmented knowledge. Unclear processes. Weak integration. Limited understanding of customer intent.
These are the real constraints.
And until they’re addressed, no platform, no matter how “best-in-class”, will deliver what it promises.
The organisations that recognise this early don’t just avoid failure, they unlock the full value of AI, because they’ve built the foundations to support it.