Healthcare organizations worldwide are rushing to use artificial intelligence to help doctors and patients, but they are hitting a major roadblock: they cannot scale these tools as fast as they want. According to a new report from Kyndryl, a major technology services company, healthcare organizations are creating too many AI experiments and pilots without being able to turn them into real, working systems. In fact, 76% of healthcare organizations report having more AI pilot projects than they can actually use in their hospitals. This is a big problem because every day, healthcare workers are busier and patients need faster care.

One of the biggest reasons hospitals cannot scale AI is that they are terrified of breaking rules and regulations. More than half (55%) of healthcare organizations are worried about keeping up with new AI rules and regulations, and only 30% feel ready to handle these new laws. To help with this challenge, Kyndryl created a new tool called "policy as code." This tool works like a computer program that takes complicated government rules and turns them into instructions that AI systems can understand and follow automatically. This means that when AI is making decisions about patient care or handling information, it automatically checks that it is following all the rules without needing a human to watch it constantly.

Healthcare systems are also using agentic AI - which means AI that can think and make decisions without a human telling it exactly what to do - in new and important ways. One of the biggest changes is that many hospitals now use AI to answer the phone when patients call with health concerns. This virtual triage works like this: when a patient calls with symptoms, an AI system talks to them first and asks questions to understand their problem, then it decides if the patient needs to see a doctor right away or if they can wait. This helps hospitals handle more patients without hiring more workers, which is very important because there is a shortage of healthcare workers worldwide. However, there is a problem: patients sometimes feel upset talking to a computer instead of a person, and they worry if the AI is giving them correct information.

Beyond answering phones, hospitals are also using agentic AI to handle boring office work. AI systems are now doing tasks like billing (sending bills to insurance companies), medical coding (writing down what doctors did in the right computer language), and writing up notes after telehealth visits where patients talk to doctors online. While these tools can save doctors time, hospitals say the results are mixed - sometimes the AI saves lots of time, but other times it does not work as well as expected. Most importantly, hospitals do not always know how many mistakes the AI is making because they do not measure errors carefully.

Another big concern for hospitals is cybersecurity and protecting patient information. As hospitals use more connected devices - like machines that check your heart, blood sugar monitors, and computers that work together - it becomes harder to keep hackers out. Many hospitals still use old machines that run outdated computer programs and cannot protect themselves from modern attacks. When hospitals add AI to these systems, it makes the problem even more complicated because AI itself can create new security problems that experts do not fully understand yet.

Healthcare leaders are also paying attention to digital therapeutic apps - special programs on phones and computers designed to help people manage diseases like depression, diabetes, or heart problems. The problem is that there are hundreds of thousands of these medical apps available, but doctors and patients do not know which ones actually work and which ones are unsafe. To solve this, a group called the IEEE Standards Association created a new list that tests medical apps carefully on 140 different points to make sure they are safe, actually work to help patients, and protect privacy.

Partnerships and conferences this week show that the world is serious about improving healthcare with AI. A big conference called Medical Korea 2026 is happening in Seoul, South Korea on March 19-22, bringing together healthcare leaders from many countries to talk about AI and how hospitals can work together across borders. Meanwhile, the University of Toronto and the Indian Institute of Science announced they are starting a new center to use AI to predict health problems before they happen, helping people stay healthy instead of waiting until they get sick.

The bottom line is that agentic AI is changing healthcare in many ways - from answering patient phone calls to doing office work to helping doctors make better decisions. However, hospitals need to figure out how to use these tools safely, keep patient information secure, follow all the rules, and make sure the AI gives correct answers that doctors can trust.