MartinJenkins is seeing an emerging trend of organisations treating AI as an ongoing experiment, with endless pilots and proofs of concept. Jarred Griffiths, our Lead for AI Transformation, argues that to move out of experiment purgatory, organisations need to approach adopting AI as an investment programme, not a trial.
In early 2024, Hutt City Council made a major decision to transform the Council through an AI investment programme. As a signal of the priority and intent, the programme was directly sponsored by the CE and led by a Tier 2 role, the Director of Strategy and Engagement. In that Director role I led the programme from the ground floor, seeing first-hand what was critical to making the AI adoption programme successful.
When we began, Hutt City Council was in much the same place as many other organisations I see now. Staff were dabbling using the tools, with 20-plus funded premium licences in place, and there was little longer-term thinking about the wider use of AI and how to scale that up. In other words, there was awareness of AI’s potential, but a big gap between that awareness and how to implement a programme that achieved tangible benefits.
As the leadership team explored its appetite for investing in AI, it was clear it wanted to shift to a higher gear. The leadership took an approach that treated the next stages of the AI adoption as a significant investment programme, to be planned for and managed much like other significant investments.
With organisations now under more pressure to adopt AI faster and in a more transformational way, I believe laying out a structured AI investment programme offers organisations and leaders a useful model for taking a more intentional path.
AI is new, but the fundamentals of successful projects are not
Most of us will be familiar with the ingredients for running a successful project – a strong investment case, clear objectives, visible sponsorship from the leadership, realistic resourcing, good governance, active risk management, and a practical plan for achieving change. These aren’t revolutionary ideas.
With the pace of AI, however, we may be at risk of forgetting these fundamentals. The pressure to “do something” with AI means we’re moving quickly to activate licensing and get pilots up and running, but we’re often not thinking through how to set projects up for success.
At Hutt City Council, having a clear investment case was central to the success of the adoption programme. The investment case set out why AI mattered strategically, what benefits we were targeting, the investment required, and how we expected that investment to translate into benefits.
A clear investment case meant the whole organisation understood where we were headed
The investment case got the executive leadership team and the wider project team heading in the same direction. As a result, the wider organisation was also clear on what we were doing. Anyone could read the investment case and understand exactly what the intention was and how it would be achieved.
The message that the staff got as a result wasn’t “We’re trialling tools”. It was:
“We’re rolling out 150 licences across the organisation initially, and developing a range of custom AI assistants. Over time we expect to see productivity improvements that could change the shape of some roles and of how work is done.”
That clarity was only possible because we took the time upfront to think through our AI programme, rather than making it up along the way.
.jpg)
Taking time to plan isn’t an excuse to move more slowly
The point is not to add process for its own sake. It’s to do enough thinking upfront so that the programme can move faster once it gets started. This avoids getting stuck in uncertainty, with shifting decision points and repeated debates about what’s being tested and why.
For Hutt City Council, we aimed to do the work on the investment case in six to 10 weeks, including use-case discovery, internal engagement, and then drafting and approving the investment case. Our investment case considered three “buckets” of activity: investing in AI licences for staff, creating a suite of AI agents, and investigating three end-to-end AI process transformations. Crucially, training and change management were called out early as being required to underpin this work.
To stay disciplined and meet that schedule, we set hard constraints upfront. We were clear about the format of the investment case – short and sharp, with minimal design – and we focussed people’s effort on the decisions that mattered and that leaders needed to consider.
Look beyond the tools to the change work needed from the whole organisation
It would be a mistake to plan AI investment around licensing or consumptions costs alone. Instead, it needs to focus on the change work required of the whole organisation.
Licences are the easiest number to find and the easiest to defend, but they are only a fraction of what it actually takes to get value out of AI. A realistic investment case needs to account for the people, time, training, and change effort required to turn AI investment into something that transforms how work is done.
For every dollar spent on licensing, it wouldn’t be unreasonable to budget an additional $1 to $2 on activity that helps achieve the benefits of these tools.
Implementation is usually the part that gets squeezed – it’s the line in the budget that looks optional when money is tight. And that’s almost always the reason programmes underdeliver.
Consider your management controls early on to ensure you can achieve benefits
Investment planning isn’t a set-and-forget exercise. Organisations can’t just approve the funding and then take their eye off the ball. That’s especially true with AI: the benefits don’t arrive on their own, and the gap between what was promised and what gets delivered can turn out to be wide.
One of the things that drove the success of Hutt City Council’s AI programme was putting tight controls around managing the benefits. We were clear about what benefits we were chasing, who owned them, when we expected to see them, and how we would measure them.
Our investment case set a specific productivity uplift target (at least 1 hour of additional productivity per staff member per day), which then drove the investment in staff capability to ensure we met this. We then committed to materialising some of these savings as roles became vacant.
We found that those controls kept us relentlessly focussed, day in and day out, on achieving what we set out to do. They were baked into how the programme ran rather than bolted on at reporting time, and they gave us confidence that we would deliver what the investment case promised.
Benefits management is often an immature capability in organisations, but it can’t be with AI. It needs to be a focus from the outset, otherwise the risk is a lot of money invested with very little pay-off.
