Where to Start with AI Adoption

So where do we actually begin?

With all this potential, it's easy to get overwhelmed and distracted by the tools.

But real AI success doesn't start with technology.

It starts with your business, your pain points, your domain expertise.

Ask yourself, where are we losing time?

Where are the bottlenecks?

What's frustrating your team or clients?

That's where the opportunity lies.

Don't start with AI.

Start with a problem worth solving.

Refine it down to something specific, something measurable.

Then ask, what would a better outcome look like?

What kind of time or cost savings could we create and how would we know if it's working?

From there, begin internally on a narrow business problem.

Aim to get the basics right, involve your team, test, and iterate.

Once that's solid, then you can scale outward.

This is the recommended progression, crawl, walk, run.

To determine the return on investment of an AI project, we first consider three core elements.

First, the true cost of the role, including salary, benefits, and equipment, for example.

Second, the annualised hours saved.

And third, the opportunity value created by redeploying that time.

An example is a marine engineering contractor performing site visits and proposals.

In the traditional process, the contractor spends four hours on site plus travel and about seven hours preparing the proposal.

The total is roughly 11 hours per opportunity with an AI enabled process.

Onsite work remains four hours that the administrative and drafting tasks are reduced.

Automated transcription captures notes instantly.

Proposal generation is cut, saving around three and a half hours per proposal.

In this example, the year one cost savings consider an average engineering rate, including overhead, at $80 per hour.

At 3.5 hours save per job across 60 projects a year, the direct saving is $16,800 annually.

After accounting for a $10,000 implementation cost, the first year ROI from efficiency alone is roughly 68%.

Plus the opportunity revenue gain.

By reducing admin and proposal prep time, capacity increases from 60 to 90 projects per year - an additional 30 jobs.

Assuming an average project value of 150,000 and a 15% margins, the added profit is approximately 675,000 per year.

Combined with the time saving benefit, the total annual gain is about $691,800 Relative to a $10,000 implementation, that equates to an annualised ROI of roughly 6800%.

Assuming steady demand and similar margins, the cumulative gain reaches nearly $3.46 million over five years, producing a five year ROI of around 34,000%.

This excludes additional downstream benefits such as faster troubleshooting on site and reduced time spent on future jobs.

You can see in this example how a small investment can unlock significant gains.

If you picture just one process in your own business, one task that consistently drains time and apply the same principle, the scale of impact becomes immediately tangible.

So once you've defined your problem worth solving and metrics for success, the next step is actually a step backward so we can move forwards with confidence.

Most of us, if we're honest, have been duct-taping systems together just to keep up with technology in life.

Over time, that leads to tech sprawl, messy processes, double handling and workflows that make you want to cry.

And if we plug in AI at this point - before we've cleaned house - what we'll get is amplified chaos.

So when I say foundations, I don't just mean auditing your tech stack.

I mean stepping back and looking at your business processes holistically.

What tools are you using?

Are they talking to each other?

What's your single source of truth?

Is it SharePoint, your CRM, your contracts or procurement system?

What tools can be consolidated?

What needs to go?

Once you've clarified that, the next step is to turn to your workflows.

Some of you may already have this documented, and that's great.

Others might find it helpful to start small.

Pick a common document, maybe a work order or invoice, and trace its journey.

Who touches it?

What tools are used?

Where does it get stuck?

This gives you a real picture of how work gets done and what needs to change.

Because only once your workflows are clearly defined and proven to work, then it makes sense to add automation and ai.

And if you find that a process needs to be reworked to align with your systems or your chosen source of truth, do that first.

Test it, use it, get it working.

Then plug in the AI.

Now that we've taken a step backward to shore up our foundations, now I'd like to take one more step backwards to ground zero.

Recently, in conversation with my IT provider, he shared a case within a local company where an employee had accessed the full company's payroll data and financial information simply by asking co-pilot.

They had not set up permissions correctly, which enabled the local tool to access folders that the employee should not have had access to.

Whether inside your system, or by experimenting outside of it, employees are already using AI tools at three times the rate leaders think they are, according to McKinsey's 2025 report.

And we all know that abstinence doesn't work.

If we don't set up responsible access, people will find workarounds.

So ground zero is this: establishing an AI policy.

A policy that: clearly defines what tools are approved, sets user access based on need and role, includes rules around confidentiality and output review, ensures human oversight is always in place for critical decisions and regularly audits, usage and outcomes for bias, accuracy, and compliance.

Once that foundation is in place, once your team understand what's allowed, what's safe and what's expected, you're in a position to move forward.

Technology doesn't lead change.

People do.

No matter how powerful the tool, it's your domain expertise and leadership that will determine whether AI delivers real value.

We know that many teams are using AI behind the scenes, experimenting without formal guidance, and as we've seen, they're asking for support to use it well and to use it safely.

This early momentum is being driven by millennial managers, those aged 35 to 44, who are some of the most confident adopters in the workforce today.

If you want AI to succeed in your business, you need people who are willing to lean in.

People who know your operations, see the pain points and are ready to lead from the inside.

Your role as a leader is to back them.

Set the tone.

Create the space.

Fostering a culture of confident experimentation.

Investing in practical training.

Encouraging experimentation, and celebrating early success.

It also means setting the tone by engaging with AI yourself and showing your team that it's safe to explore, as well as shifting the narrative from AI replacing to AI augmenting human capabilities, freeing up time for higher value work.