In Priority One’s recent Business Sentiment Survey, 85% of businesses identified AI or automation as likely to impact their organisation over the next 12 months. So as AI shifts from a “should we?” to a “how do we?” conversation, businesses are grappling with where it fits, what it delivers and how to make it stick. We speak to Simon Thomas, AI Principal at Rush Digital to understand what’s changing, where the most value lies and how to keep making progress.

Have you noted any changes in the last 12–18 months in terms of how businesses are approaching AI? 
Eighteen months ago, most conversations started with “should we be doing something about AI?” Now it’s, “we know we need to – but where do we start?” 

That’s progress, but it’s created a new kind of paralysis. There’s so much noise. Every week there’s a new tool, model or think-piece declaring everything has changed again. People are already fully stretched doing their actual job, so expecting them to also sift through an overwhelming feed of AI content and make confident strategic decisions is unrealistic. 

What are the main reasons businesses are investing in AI right now? Is it about cost savings, growth, staff shortages or something else? 
Capacity is probably the biggest driver – not just cost savings, though that’s part of it. There’s pressure to do more without proportionally growing headcount. Tight margins, staff shortages, increasing complexity. AI gets positioned as a way to close that gap. Growth comes into it too, but usually later, once the basics are working. 

What I’d push back on is the idea that AI is a simple plug-in fix. ‘Give people the tools and they will come’ is not a strategy. The best results come when it’s treated as a people investment as much as a technology one. 

Are you seeing large organisations getting the most benefit or are there good starting points for small businesses? 
Both have strong entry points, but the dynamics differ. Large organisations have budget and mandate, but also bureaucracy, security reviews and competing priorities. They can also fall into the trap of thinking one team can represent the whole organisation and produce a single AI roadmap. 

Leadership has a distinct role, but the people doing the work are the ones who best understand where AI can help operationally. 

Across all size businesses, governance, security and privacy matter from day one. Shadow AI – people quietly using unapproved tools with company data – can happen anywhere. It’s not about restriction; it’s about putting guardrails in place before the habit forms. 

A motivated internal champion, leadership backing and a clear framework that allows safe experimentation – that combination moves things regardless of size. 

What are some of the most common types of AI or automation projects you’re working on? 
We think about it as a flywheel: 

People: Building individuals who can use AI effectively in daily work. Structured programmes accelerate capability in ways ad hoc tool access can’t. The goal is people who understand what AI can do in their role, so they start spotting inefficiencies no one else would see. 

Making workflows AI-aware: Processes designed with AI in mind create clearer requirements for tooling, rather than adopting tools speculatively. 

Tooling: Solutions connected to data and workflows that improve outcomes. When this works, it expands what people believe is possible and drives demand for more capability building – feeding back into people. 

The flywheel breaks if one of these steps is missing. Many organisations overinvest in tools and skip people, so adoption flatlines. Or they run training that doesn’t connect to real processes and enthusiasm fades. The starting point depends on the organisation, but the goal is always all three working together. 

Where are businesses seeing quick wins? 
Individual productivity – drafting documents, summarising meetings, researching. It builds confidence in both the tools and people’s ability to use them. 

But the mistake is stopping there. Real value emerges when AI connects information across systems that previously didn’t talk to each other – moving from saving time to doing things that weren’t previously possible. 

What are businesses often overestimating or underestimating about AI? 
Overestimating: speed of return. “ROI within 90 days” is common, but rarely realistic. Value takes time to emerge and is hard to measure early. Typically, there’s initial excitement and adoption, then a messy middle where opportunities are identified, systems connected, data organised and processes refined. If you get through that stage, significant uplift is possible – but it requires patience. 

Underestimating: change management. The technology is rarely the hard part, and it will keep changing. The real challenge is helping people navigate uncertainty, build new habits and manage concerns about job security. 

What tends to slow projects down? 
“Our data isn’t ready” is the most common. Sometimes it’s valid, but often it becomes an ongoing excuse. The better approach is to start, identify priority opportunities, then work out what data is actually needed. 

The other is waiting for a complete AI strategy before starting. But familiarity with tools – what they do and how they behave – is a prerequisite for strategy, not the other way around. 

Where do businesses get stuck? 
After the pilot. There’s initial enthusiasm, results look promising – then nothing happens. The pilot doesn’t become a process. It usually comes down to ownership: who is responsible for making this part of how work gets done. 

Can you talk about a recent project and what it achieved? 
We ran a three-month AI capability programme with a large agricultural cooperative – 32 technical staff, scaling from early adopters to broad adoption across the team. It included an AI-readiness assessment, prompt engineering and agentic coding workshops, and sessions on tools like Claude, Cursor and GitHub Copilot. It culminated in an AI Hackathon where teams built functional prototypes in four hours that previously would have taken weeks. 

We also developed a custom AI playbook, established an internal AI Guild for peer learning, and embedded a self-sustaining adoption workflow so it didn’t end when we did. 

What changed for that business after implementing it? 
The numbers were strong: 95% using AI tools daily or constantly, 100% adoption (up from 87%), and 87% achieving one to five hours of weekly productivity gains – averaging 4.6 hours saved per person. One team member achieved a 92% reduction in recruitment task time. The hackathon had over 90% attendance and an 8.5/10 satisfaction rating, with teams building internal AI assistants and recommending AI to peers at 9.5/10. 

However, even with those results, we saw the same plateau effect. The initial uplift didn’t compound – enthusiasm settled, the day job reasserted itself, and incremental improvement didn’t happen on its own. We’re now working to extend the programme to adjacent teams – creating cross-pollination of value and ensuring capability spreads. That’s what turns a successful pilot into something structural. 

What’s one practical step a business could take in the next 3 months? 
A few things – ideally in parallel. 

First, put a basic AI policy and approved tools in place. People are already experimenting, and without a framework that happens in the dark, often with company data in unvetted tools. 

Second, map your organisation before your technology – who your people are, what they’re trying to achieve, what they do day-to-day and what they produce. Even a rough map surfaces opportunities a top-down strategy misses. 

Third, run a structured AI familiarisation programme. Not “here are the tools,” but guided exploration in the context of real work. The goal is hands-on experience so opportunities become obvious to the people doing the work. 

What advice would you give to businesses in Tauranga that are just starting out? 
Start with curiosity. Find one internal champion and give them time to explore. Get hands on tools – you don’t need perfect data, a roadmap or a dedicated AI team. And talk about it. People are already experimenting, and those stories matter. 

The businesses best positioned in two or three years won’t necessarily be the ones who picked the right tools early – they’ll be the ones who built the habit of learning and culture of sharing. 

Sign up for our fortnightly business news updates here.