Contact Us

Why AI Adoption Feels Busy but Doesn’t Stick

Organisations running in place with AI tools but not moving forward

Why AI Adoption Feels Busy but Doesn’t Stick

Why shared fluency, not more tools, is the missing layer in lasting AI capability

As I come to the end of the Certified Implementer program, it’s given me a chance to pause and reflect. Not just on the program itself, but on the same issues that keep surfacing through conversations with other implementers, leaders, and with John Munsell at INGRAIN.

What started as a practical implementation pathway has evolved into something broader. A way of understanding why so many AI initiatives struggle to move from early activity into capability that lasts and what differentiates approaches that actually stick.

Because the issue most organisations are facing right now isn’t access to AI. Or even ambition.

It’s whether the organisation can actually take it on and use it well.

My personal aha moment came while doing an INGRAIN fluency assessment. I realised personal fluency wasn’t enough. The harder question wasn’t whether individuals could use AI well, but whether the organisation could still function when decisions drifted, results degraded, or judgement went off track.

When Activity Doesn’t Turn Into Capability

Across many organisations, AI use is increasing.
 

Tools are being tried.
Pilots are running.
Outputs are being produced.

 
From the outside, it looks like progress. Inside the organisation, it often feels less certain.
 
But underneath, behaviour across the organisation hasn’t really changed. Confidence varies by individual. Outputs are inconsistent.

Leaders aren’t always sure what’s acceptable, defensible, or safe to scale.

 
This is the point where AI adoption tends to stall.
 
Not because the technology fails, but because uncertainty starts to outweigh confidence.
 

AI rarely fails loudly.

 
It tends to fade when people stop trusting how it fits into their work.

👉  “As McKinsey notes, ‘AI’s biggest hurdle isn’t tech, it’s organisational readiness.”

 

Read on to see why… AI Isn’t Moving Too Fast. Organisations Are Just Struggling to Absorb It.

 

why AI adoption often stalls

When Activity Doesn’t Turn Into Capability

Across many organisations, AI use is increasing.
 

Tools are being tried.
Pilots are running.
Outputs are being produced.

 
From the outside, it looks like progress. Inside the organisation, it often feels less certain.
 
But underneath, behaviour across the organisation hasn’t really changed. Confidence varies by individual. Outputs are inconsistent.

Leaders aren’t always sure what’s acceptable, defensible, or safe to scale.

 
This is the point where AI adoption tends to stall.
 
Not because the technology fails, but because uncertainty starts to outweigh confidence.
 

AI rarely fails loudly.

 
It tends to fade when people stop trusting how it fits into their work.

The Common Misdiagnosis

When this happens, it’s often explained away as a skills problem.

 

More training.
Better prompts.
Stronger tools.

 

Those things help, but they rarely resolve the underlying issue on their own.

 
For a while I assumed the answer was getting individuals sharper. Better at using the tools, faster at getting results. It took a few conversations with teams who were already doing that well to realise the problem was never individual capability. It was whether the organisation had any shared idea of what good looked like.

 

What’s missing in most cases is a system that converts individual AI use into shared ways of working over time, rather than remaining separate from how the organisation actually operates.

 
Without that system:
  • learning stays individual
  • good practice doesn’t spread
  • leaders struggle to govern what they don’t fully understand
 
This is why organisations can look busy with AI without becoming meaningfully better at using it.
Guardrails metaphor showing how structure accelerates safe adoption

Why Tool-Led Thinking Makes This Worse

One reason this keeps happening is that AI is still being approached primarily as a tool problem.

When organisations lead with tools:
 • experimentation fragments
 • learning stays personal
 • results are hard to explain or defend
 • governance arrives late, and often heavy

The counter-intuitive insight is that structure doesn’t slow AI adoption.

It legitimises it.

A useful metaphor is guardrails on a steep mountain road. Guardrails don’t make you drive slower. They let you move faster without constantly second-guessing every turn.

AI works the same way.

Without clear boundaries and shared expectations, people hesitate. They edge forward. Progress feels risky.

Capacity Appears, But Judgement Doesn’t Automatically Follow

AI is very good at amplifying execution.

 

It improves throughput.
Reduces friction.
Speeds up production.

 
In organisational terms, it strengthens execution and process.

 

What it doesn’t automatically strengthen is judgement and alignment.

 

This is why capacity appears, but without deliberate direction it gets absorbed back into busyness. More activity. More motion. Not necessarily better decisions being made.

 
This shows up repeatedly when organisations skip the work of clarifying intent, decision rights, and success criteria. Rarely treated as part of how the organisation actually operates.

The Missing Layer: Shared Fluency

The organisations that move past this stage don’t just train people to use AI.

 

They build shared fluency.

 
Shared fluency isn’t about everyone knowing the same prompts or tools. It’s about having a common understanding of:
  • what AI is for
  • where it’s appropriate to use
  • how outputs are judged
  • when human judgement steps back in

 

This is where concepts like scalable prompting actually matter. Not as clever prompt design, but as a way of embedding judgement, standards, and intent into repeatable practice.

 

Without shared fluency, scale feels risky. With it, confidence grows.

 
This is not a training gap.
It’s a systems gap.
Visual of building shared understanding in AI use

Why Tool-Led Thinking Makes This Worse

One reason this keeps happening is that AI is still being approached primarily as a tool problem.

 
When organisations lead with tools:
 • experimentation fragments
 • learning stays personal
 • results are hard to explain or defend
 • governance arrives late, and often heavy

 

The counter-intuitive insight is that structure doesn’t slow AI adoption.

 

It legitimises it.

 
A useful metaphor is guardrails on a steep mountain road. Guardrails don’t make you drive slower. They let you move faster without constantly second-guessing every turn.
 
AI works the same way.
 
Without clear boundaries and shared expectations, people hesitate. They edge forward. Progress feels risky.

👉  “Research from BCG reinforces this: shared AI fluency increases transformation success by up to 30%.”

Progression Matters More Than Adoption

AI capability develops in stages.

 

Early experimentation has value. But without progression, it plateaus.

 

This is why maturity frameworks like the AI Mastery Ladder exist. Not to label organisations, but to help ensure that learning, application, and governance evolve together rather than at random.

 
When progression is explicit, capability builds over time. When it isn’t, organisations keep resetting.

👉 Read more… Still stuck at the Explorer stage? Why AI change requires more than curiosity

 

why progression matters more than early enthusiasm

Why This Changes How We Think About Implementation

So the question becomes: who holds this?

 

Not the technology. Not a training programme. Not a one-off project team.

 

Someone needs to sit at the intersection of learning, governance, and day-to-day use. Someone whose job isn’t to build AI capability in isolation, but to make sure it moves through the organisation in a way that sticks.

 

That’s what a Certified Implementer does.

 

Not as a trainer. Not as a tool expert. And not as someone doing AI for the organisation.

 
In practice, the role exists to: 
  • hold the sequence of adoption
  • turn individual learning into shared practice
  • introduce structure early enough to enable speed
  • prevent premature automation and reactive governance

 

INGRAIN provides the operating model that supports this. It treats AI adoption as a shared way of working rather than a collection of tools, prompts, or courses.

 
Tools help people do more. Systems help organisations decide better.
 
 

Why This Matters Now

AI capability is compounding faster than most organisations can comfortably absorb.

 

That gap doesn’t resolve itself.

 

Without a clear operating model, organisations keep investing in activity that resets each cycle.

 
Even with structure, some organisations still resist slowing down long enough to decide what they’re actually optimising for.
 
The organisations that get the most value from AI won’t be the ones moving fastest. They’ll be the ones building the conditions that let progress stick.
 
 

Final note

This is the thinking that shapes how we work at Dovetail.

 

We follow the INGRAIN methodology to help organisations operationalise AI use and move beyond activity into durable capability, in a calm and deliberate way.

 
If this way of thinking reflects what you’re seeing in your organisation, we’re open to a conversation. Not to sell tools, but to help leaders get clearer about what to do next and what can wait.
Curious? Let's take a look together

Frequently Asked Questions

(FAQs) for Australian Businesses

Q1. Why do many AI initiatives fail to scale?
Most AI initiatives don’t stall because the technology fails. They stall because early experimentation isn’t converted into shared organisational capability.

Individuals learn quickly, but learning stays local. Confidence varies by team. Outputs are hard to explain or defend. Leaders struggle to govern what they don’t fully understand. Over time, uncertainty outweighs momentum.

Scaling requires systems that align learning, decision-making, and oversight so early gains compound instead of resetting.
AI training improves individual skills: using tools, writing prompts, automating tasks.

AI capability is organisational. It shows up when AI use is consistent, trusted, explainable, and repeatable across teams. That requires shared understanding of what AI is for, where it’s appropriate, and how outputs are judged.

Training is a necessary input, but without systems that turn learning into shared practice, it doesn’t translate into lasting capability.
AI adoption isn’t a one-off project. It’s a sequence.

As AI use increases, organisations face recurring questions: what’s acceptable, what’s defensible, what should scale, and who decides. Without someone holding that sequence, learning stays individual and governance tends to arrive late.

A Certified Implementer helps turn experimentation into shared practice. They introduce structure early enough to build confidence, and help ensure skills, workflows, and governance evolve together rather than at random.

The role isn’t about control. It’s about making AI use normal, trusted, and sustainable.
INGRAIN treats AI adoption as a capability problem, not a tool or training problem.

Instead of starting with what to build, it starts with questions of judgement and intent: what decisions are being delegated, what must remain human, and what needs to be true before scaling.

It provides an operating model that sequences learning, use, and governance together so AI doesn’t depend on a few confident individuals, and doesn’t trigger reactive oversight later.
The roadmap treats AI adoption as a progression rather than a project.

It sequences strategy, skills, governance, and culture so they mature together. Intent and boundaries are clarified early, shared fluency is built across leaders and teams, and governance evolves alongside real use rather than being added after problems appear.

The aim is to make progress deliberate and cumulative, not accidental.
In practice, no.

Most AI adoption stalls because people lack shared expectations. They’re unsure what’s acceptable, how outputs will be judged, or whether their work will hold up to scrutiny.

Clear boundaries and decision rights reduce uncertainty. Like guardrails, structure allows teams to move faster without constantly second-guessing themselves. It doesn’t restrict progress. It makes it sustainable.
On its own, AI training isn’t particularly expensive. The real cost comes when training doesn’t change behaviour or translate into organisational capability.

Good training often creates excess capacity. The return depends on whether the organisation has systems to reinvest that capacity into better decisions, improved workflows, and reduced rework.

The question isn’t the cost of training. It’s whether learning is converted into durable return.

Still feeling stuck? You’re not alone but you don’t have to figure it out solo. Our DMA helps you cut through the noise and focus on what matters most.
Get Started With Our Digital Maturity Assessment Today!

Share this post with your friends.

Facebook-f Linkedin-in Instagram
Agency Arcade, About Us - Agency Arcade, Contact Us - Agency Arcade, Our Services - Agency Arcade