In healthcare, artificial intelligence (AI) only delivers cost savings when organisations reinvent their core processes. This doesn't mean rolling out AI broadly. It means making sharp choices, says PwC expert Nicole Lentink.
Most healthcare executives view AI as a means to streamline processes. Automate the administration, ease the burden on doctors, do more with less and costs fall. It sounds logical. In practice, it isn't. And it often fails.
When you apply AI to your existing way of working, it drives costs. You save on a task, but the structure around it remains: the processes, the care activities, the roles, the locations, the overhead, the way you're paid. You pay for the technology and for the model you won't let go of. AI isn't your strategy. Healthcare transformation is the strategy. AI is the most important means to free up healthcare capacity, so you can deliver better care.
Put healthcare transformation at the centre
Look at where AI lands in healthcare today: decision aids, administrative support, patient communication and planning. AI reporting assistants automatically convert consultations into reports, delivering modest efficiency gains on defined tasks. Doctors and nurses reclaim time, reducing their workload. That matters. But this doesn't create substantial value through reduced costs and better accessibility.
The challenge? We continue delivering care as we always have. We know appropriate care is necessary, but achieving it is complex. The risk is that separate use cases don't fundamentally change how we work or only tackle bottlenecks incrementally. Worse, when AI makes care more accessible for patients, it can further increase demand for care. Healthcare transformation must be central. AI is the means to achieve it.
More AI isn't the answer, sharper deployment choices are
The executives who make AI work choose decisively. Where it works, we see the same pattern:
- Analyse where impact and feasibility are greatest, choose a handful of areas to truly transform and deliberately leave the rest. Dare to redesign the back office. There, impact is quickly measurable, scalable and realisable due to lower risk and proven applications beyond healthcare. The real savings you achieve here create space for care transformation.
- Moving from separate use cases to structural application requires building a solid foundation: governance, compliance, IT capacity, data and execution power. Equally important is giving employees space to understand AI, practise with it and become enthusiastic. Organise that space with focus. Let experiments connect to chosen transformation areas, work within clear frameworks and determine upfront when you scale up or stop an application. You'll stimulate adoption without falling into non-committal, bottom-up pilots.
- Impact doesn't come from more AI tools. It comes from fewer. From a few places where you genuinely reinvent your business model. From relocating services, adjusting functions and reorganising locations. You only deliver on AI's big promises when you change your business model. Not when you stack technology on top of your current system.
AI's impact lies in what you release and transform
AI benefits are diffuse, often only visible long-term and difficult to attribute to one department. Better quality, fewer errors and more time for complex care are all results, but rarely visible in one budget line. So why make painful choices now based on benefits you can't pinpoint?
What is measurable is what you scrap or change. A service you no longer deliver, a staffing level that shifts, a location that's reorganised, a care activity that disappears. The hard impact isn't on the tool. It's in what you dare to release and transform. And you only release when you consciously choose. Not choosing isn't a safe route. It's the route where costs mount until someone else makes the choice for you.
Choices you as an executive can make now
We won't end with a call to 'be strategic with AI'. You've heard that enough. Instead, three concrete choices:
- Identify and select the most promising core processes based on impact and feasibility. Focus on one to three processes you'll truly redesign, not just automate tasks. You can't do ten simultaneously.
- Name what you're releasing: a service, a location, a working method. No redesign without scrapping something.
- Measure your AI deployment against your cost base and added value, not against the number of tools you've rolled out.
Affordability doesn't come from more AI. It's the result of sharper choices and the courage to let something go. The question isn't how much AI you roll out. The question is what you dare to scrap.
Decode your AI value and reframe your approach to reinvention and long‑term growth
Read how CZ and Santeon are making AI work at scale (in Dutch)
About the author
