What I’m Seeing in Enterprise AI — And What Actually Works

Most organisations aren't debating whether to use AI anymore. The conversation has shifted to execution — how to implement it effectively and see meaningful results quickly.

In my role, I spend a lot of time talking to customers about their AI journey. The pattern is consistent. The common assumption is that progress stalls because the technology isn't ready. In reality, it's usually because changing habits is harder than installing software.

Start with the work, not the tech

The best advice I give customers is simple: don't start with AI — start with the work.

The organisations making progress aren't asking, "What can AI do?" They're asking, "Where are we wasting time, effort, or energy today?"

The answers are usually obvious once you ask the question. Too much time summarising meetings. Emails and documents that take far longer to draft than they should. Information scattered across tools that nobody can find when they need it. Leaders who can't get a clear picture quickly enough to act on it. When AI is applied to problems people actually feel, value shows up fast — and confidence follows.

But even when the capability is there, progress depends on whether people are willing to work differently. People default to what's familiar. They draft the same way, run meetings the same way, search the same way. AI doesn't fail because it can't perform. It fails because no one changes how they work.

Getting AI over the line internally

When organisations struggle to move forward, it's rarely a technology problem. It's usually that different groups are pulling in different directions at the same time. IT is focused on security and risk. Leaders want ROI. Employees want to know what's in it for them personally. Each group is reasonable on its own — but without someone connecting the dots, nothing moves.

That's a big part of what our role looks like in practice: showing AI working inside the tools and scenarios people actually use, translating capability into business outcomes rather than buzzwords, and helping leaders articulate clearly why AI is being adopted and how it will be governed.

But alignment alone isn't enough. There's another layer that often gets overlooked — reinforcing new working habits. Adoption isn't a launch event. It's repetition, small wins, and those wins compounding over time. One-and-done training rarely changes behaviour. Without reinforcement, people revert. Once someone genuinely experiences AI improving their day-to-day work — and that experience is consistently reinforced — momentum follows naturally.

Does ROI matter?

Yes — but not always in the way people expect.

Early on, ROI isn't just financial. It shows up in time saved, less rework, faster turnaround, better-quality outputs, and staff who are less frustrated with how their day runs. Those things matter before a spreadsheet can prove them. And they compound — financial ROI follows consistent use, not the other way around.

If organisations wait for a perfect ROI model before starting, they tend to wait too long. The return isn't created by buying AI. It's created by using it consistently enough that new habits actually form.

Choosing AI in an enterprise environment

Capability is part of the conversation — but it's rarely what decides things. In enterprise environments, the questions that actually matter are around security, compliance, data protection, how it integrates with existing systems, and who owns what. Even highly capable AI will underdeliver if it isn't embedded into the workflows people use every day. In practice, success comes down to trust, integration, and consistent use.

Governance doesn't have to slow you down

Good governance enables AI. Poor governance kills it.

A strong starting point is knowing where your data lives, being clear on what AI can and can't access, and defining ownership and accountability before something goes wrong rather than after. The key is keeping governance practical and human. If it's overly complex, it won't stick — and governance that nobody follows isn't governance at all. It should create confidence, not friction.

The best AI strategies I've seen

The strongest approaches don't treat AI as a standalone initiative or a project with a finish line. They focus on clear business outcomes, use cases that are specific to roles and responsibilities, leadership that's genuinely capable rather than just supportive in name, and a commitment to measuring and iterating rather than declaring victory at launch.

When that becomes part of how work actually gets done, AI stops feeling like a risk or an obligation. It just becomes the way things work.

Bottom line

The organisations winning with AI aren't the most technical. They're the ones who were honest about where work was breaking down, deliberate about how they introduced change, and patient enough to let new habits take hold.

The real shift was never the technology. It's how people choose to work.