In this next episode of The Heidrick & Struggles Leadership Podcast, Heidrick & Struggles’ Federico Guerreschi speaks to José Pedro Almeida, a top health AI and chief AI strategist and advisory board member at Intelligence Ventures, an emerging venture capital firm dedicated to cultivating innovation at the intersection of artificial intelligence and healthcare within the United States. In this interview, Almeida discusses the potential of generative AI and its impact on businesses and healthcare organizations worldwide, explaining how AI-driven digital workforce models can revolutionize patient care and operational efficiency. He also emphasizes the importance of collaborative efforts among technology companies, healthcare leaders, and stakeholders to navigate the complexities of AI integration.

Below is a full transcript of the episode, which has been lightly edited for clarity.

Welcome to The Heidrick & Struggles Leadership Podcast. Heidrick is the premier global provider of senior-level executive search and leadership consulting services. Diversity and inclusion, leading through tumultuous times, and building thriving teams and organizations are among the core issues we talk with leaders about every day, including in our podcasts. Thank you for joining the conversation. 

Federico Guerreschi: Hi, I’m Federico Guerreschi, principal in Heidrick & Struggles’ Milan office and a member of the global Technology Services Practice, which focuses on the technology, digital, and innovation space across a wide range of industries. In today’s podcast, I’m excited to speak to José Pedro Almeida. JP has been chief AI strategist at various companies and is now an advisory board member for Intelligence Ventures, an emerging US venture capital firm investing at the intersection of AI and healthcare within the United States. JP, you are recognized as one of the world’s top 70 health AI brains for your work on some of the world’s most transformative big data and artificial intelligence platforms in large healthcare organizations. You have also recently been named among the top 25 digital health leaders to follow in 2024. JP, welcome and thank you for taking the time to speak with us today. 

José Pedro Almeida: Pleasure to be here, Federico. 

Federico Guerreschi: JP, could you please start by telling us why generative AI is a real game changer within the technology sector?

José Pedro Almeida: We’re moving from an age where we had to explicitly tell computers how to do things, they were merely tools for us, to an age where computers have started understanding our world. That is highly transformational. And that has happened because of a breakthrough in a specific area of AI: deep learning. Deep learning has a new architecture, called transformer architecture, that suddenly allowed these models to be trained at internet scale. And when you train models at internet scale with all the information that’s available publicly from forums, books, and so on, these models start to show some emerging capabilities. They start being able to summarize any text, to translate to any language, and to [draw conclusions] on top of the data that they learn from because they start to be world models—they can predict what is happening next. So we are talking about a technology that suddenly starts to become indistinguishable from humans because these models, as they learn and understand our world, will start to hear, to speak, and to act as we do.

Let me give you an example of how this will be impactful on any company. For several years, Tesla had their fully self-driving cars running on several AI models, and they were trying to take this forward. But there were 300,000 lines of code telling the car that when it saw a stop sign, it needed to stop. This was an if–then–else approach. Now, the latest fully self-driving cars that Tesla is testing are running end to end on neural networks, which means that they only capture the images from all the fleet of their cars, and then the network is trained on that scale. And the car suddenly starts to drive; it starts to imitate humans. You can build on this use case to see what’s going to happen. Any company needs to pivot fast to this kind of approach. 

Federico Guerreschi: Since generative AI is a game changer for all companies, what strategic vision should businesses take to embrace this technology and gain a sustained competitive advantage?

José Pedro Almeida: The first thing is for CEOs and executive teams to realize how fast-paced this technology is, that the traditional way of doing things is coming to an end. This is unstoppable. The pace of generative AI is not the same as cloud or mobile, where you could wait three to five years to pivot. This technology is evolving at 10 times the pace of Moore’s law. So, you need to start thinking about rewiring your organization, because if you don’t, you will become obsolete. This will create such a digital divide between the companies that transform themselves and those that do not that those who do not will get left behind. 

So, I think executive teams need to approach this from two angles. One is by adopting what I would call everyday AI, or augmented AI, basically leveraging all the copilots that are already available in, for instance, the Microsoft or Google stack, which are quite fast to deploy and can give you quick wins. And infuse these copilots in your entire organization. We are talking about simple things like using copilots to help write and summarize emails and threads of emails, to help your sales team write a proposal, or even having AI agents listen to all the calls inside the company, the thousands of meetings that happen worldwide in each company, and summarize what happened in that call in a to-do list and send it to every director and member who was in that meeting. Just these small improvements are huge in these large organizations. And so I think this is the first angle that you need to tackle.

Also, how do you empower and augment your frontline workers? You do this by leveraging the data that is already in your IT systems. When employees are in front of a customer, for instance, these models can summarize what happened in the customer’s last four visits and bring that information up in a flash of a second. These models are really good at doing that. So, instead of the frontline worker having to have in his head 600 document pages, he can have a copilot that helps him every day in his work.

The second angle is what I would call game-changing AI. This means completely rethinking how processes can be done with a technology that fundamentally shifts how humans and machines interact. Take, for instance, a large health insurance company with 500 call center operators. More than 80% of the calls are for simple things, like a customer calling to see if his insurance policy is active or if he has a certain coverage. These types of calls can now be offloaded to AI agents. The AI agents will speak the same way as humans do, and they will even have the same breathing patterns as humans do while they are speaking. They can have an Irish accent if an Irish customer is calling. And this will offload 80% of the demand that is coming to the call center. But more than that, an AI agent can do this in parallel. So, it can answer 50, 60, 100 customers in parallel, which is something not possible today. 

To give you another example, imagine your marketing department generating 10,000 birthday emails for each of your customers, with a personalized video from your CEO offering congratulations for their birthday and thanking them for buying specific products from your company. And the CEO has never filmed that video; it was all AI generated. This is possible to do today, and if you don’t realize what is happening, you are lagging behind. 

I will end by giving an example that just went public. Tyler Perry, a well-known actor, canceled an $800 million contract to build a studio because he saw what OpenAI and their text-to-video models were doing. Any company needs to take the same approach, have the same mindset, because this will impact operating expenditures and capital expenditures at the same time. 

Federico Guerreschi: Moving to people, what new leadership skills and capabilities will companies need to truly become data- and gen AI–driven and to make it successful across the enterprise? 

José Pedro Almeida: Most executives underestimate how hard it is to do a broad-scale transformation inside the company. There’s a big distance between hype and execution, which leads to 80–90% failure. And that’s because they usually do plug-and-play things that they buy from a certain vendor or a consultancy company. But in the longer run, that doesn’t work. And that doesn’t work because you need to invest time in getting the foundations right. And if there’s one thing my 15 years of experience inside these large organizations has taught me is that the first thing that you need to account for is the organizational setup. And I would argue that the most important thing is the full commitment from the executive team in the transformation, and that starts with the CEO. 

But at the same time, the members of the executive team are super busy—they are looking at the P&L, they are looking at quarterly earnings—and they don’t have time for AI. And what often happens is that they offload this to IT, which is also overloaded with cybersecurity and resilience and all of that. So, what I think needs to happen is that you need to have an executive team member, which is the chief AI officer, who is thinking about these issues every day, who is thinking about how he is going to rewire the company and redesign the company in each workflow, from legal to marketing to sales. And this member will influence not only the senior leadership team but also the entire company. Because when you have a chief AI officer, you are sending a message to the organization that this topic is very important for the company. 

So, I would say that is the third pillar: how do you empower the transformation? And that clearly comes from the personal involvement of the CEO. But that said, this is not an occasional pronouncement; this is something that you need to do every week, every month, sending that message, because that will put this on the agenda for everyone, for every director, and that will clear the road for the transformation. Again, we are talking about something that will fundamentally shift the way people work with technology, and so that requires a lot of change management. If the CEO is not empowering this every day, the transformation will fail.

And the second angle, besides the organizational foundation, is the technology foundation that you need to build. And from my experience, the first thing you need to do is to decouple the data and AI transformation from the IT transformation. That means you need to have a separate data platform that is capturing the information from all the systems every day. You leave the IT running the operational systems; you don’t interfere with that. But on top of this platform, you are going to build the intelligence layer of the company. And that means that, workflow by workflow, you are going to plug these AIs on top of this platform, and you are starting to help, in real time, each business unit by giving them the power of this intelligence. And I would argue that it is not only what the AI can do to boost the productivity of each employee, but it is also the capability that you suddenly gain to capture value from your customers. 

Just imagine that this platform, when it is horizontally deployed across the company, will capture all the phone calls, all the emails, all the WhatsApps that would exchange with customers. And you, in a flash of a second, can leverage insights from those customers. You can ask questions like “What are customers saying about my company?” and “What are they complaining about?” by cross-analyzing all these formats, because these models are multimodal; they can extract value from text, images, sound, and video. And that’s a fundamental shift in the way you’ll see these models being deployed. 

Federico Guerreschi: Shifting the conversation to healthcare, why is generative AI a unique opportunity for the sector?

José Pedro Almeida: First, because 80% of healthcare information is unstructured, meaning it is contained in clinical notes, medical images, pathology reports, and so forth. And further, because 97% of the data that is produced annually by any healthcare organization is not used at scale. So, you have a unique opportunity to transform a sector that is lagging behind and to transform this data into knowledge. The reason it has not transformed yet is because this is a high-stakes business. You cannot stop a healthcare organization or a hospital in order to change all the IT systems; it is not possible. And so that’s the first thing.

The second thing you need to account for in healthcare is a workforce problem. After the pandemic, the workforce dropped more than 30%. That means you cannot scale a doctor to every patient and every bed. So, you need new intelligence approaches, what I would call a digital workforce that complements human labor, that acts every day, working side by side with clinicians, and allows them to perform at the top of their license. 

My background in healthcare has shown me that there are a lot of small issues that happen every day: miscommunication among teams, young doctors, senior doctors, late nights, and weekends. The level of care that is being provided is not equal in each region and in each healthcare organization. You have an opportunity with these AI models to bring a standardized level of care, because these are tireless models that work 24/7 alongside clinicians. 

To finish up this point, I think there’s a huge transformation that is happening in healthcare with these models. We are moving from a previous world where we had what I would call narrow and traditional AI, which was AI that would only work in a specific scenario for a specific problem, to one with generative AI, which understands the context and can analyze clinical notes, medical images, the sound of certain equipment, all at the same time, and produce an inference on top of that. And that closely approximates how a doctor makes decisions. When a doctor is looking for a nodule in a CT scan, he will not only look at the image, which is what these current AI models do, but he will also look at the clinical history of that patient, he will look at the blood results, he will look at the pathology report, and based on all of that, he will produce an inference. So, these models are starting to be very close to the way we make decisions and so will become much more helpful to clinicians. 

Federico Guerreschi: How is generative AI challenging the traditional deference to medical judgment, and what steps should healthcare leaders take to reshape the culture and fully leverage artificial intelligence? 

José Pedro Almeida: I think this is not only an issue for healthcare leaders. We need to understand that healthcare is complex, it’s a high-stakes business, and it involves multiple stakeholders. So, every stakeholder has a role to play to take AI and gen AI forward in healthcare and to change these dynamics. 

There are three stakeholders. The first stakeholder are the tech companies. Tech companies have been selling the wrong message for AI in healthcare, and that’s one of the reasons why it has not had much progress. They have been selling an AI that will replace doctors, and that immediately kills your transformation. It’s too naive to think that even these powerful models will replace a doctor. If you’ve never worked with a doctor before, you can imagine this. But if you work side by side with doctors, you understand that there’s a part of their reasoning that you might call a magic art that cannot be explained. I guarantee that if you have a serious disease, you will go to a doctor. So, what I think technology companies need to do is focus on the two-thirds of the medical time that is spent on back-office tasks, on mundane tasks that are not related to a diagnosis, and that can be automated. For instance, summarizing all the history of a patient or encapsulating the doctor-to-patient relationship in an outpatient visit by capturing what was said automatically and not having to write it all down on the computer. These are the areas where technology companies need to focus. 

Then the healthcare leaders, the executives who are managing these organizations, also have a role to play. They need to think much more on a computational care level, about how the computers in their organizations can help their clinical teams. I can give you an example from a large project that I led, where for a public hospital with 1,000 beds we built a huge foundational platform that analyzed 500 billion data points. It predicted patients who were going into the ICU, and it would summarize clinical notes for doctors. And this was done back in 2015. The executive team formed multidisciplinary teams between the doctors and the tech teams, and this synergy was really important because it brought trust. They were forming teams with young doctors who then influenced the senior ones, and that’s how word spread in the organization. Suddenly you started to see new ideas popping up in the clinicians’ heads, because the senior ones started seeing the younger ones use these tools, and they understood that the younger ones had a new tool set, and they also wanted to be part of that. And that’s how you change the culture in an organization. 

And, finally, in terms of the dynamics in the interaction with patients, I can say from my perspective that it will completely change how doctors and patients relate to each other. What you will see in two to three years’ time is a [highly] informed patient with the best medical knowledge at their fingertips, on their mobile phone. These models are coming to the edge and are coming to your mobile phone. So, the patient will be able to discuss his clinical case with a doctor at a level of knowledge that was never possible before. Doctors will need to prepare up front and see these copilots as personal assistants of patients. You can also imagine a world where the doctors will prescribe these models to be personal assistants to patients with certain diseases. So, there are a lot of dynamics that will change. 

Federico Guerreschi: Can you highlight some more specific scenarios where you see generative AI being deployed in the healthcare sector? 

José Pedro Almeida: Thinking from an executive perspective, once again we need to think about this digital workforce, which I would call a super-staffing workforce, where we start to have AI agents, models that can act on our behalf, impacting every part of the patient journey. I would say there are three parts to this: before a visit, during the visit to a healthcare organization, and aftercare. 

Before a visit, these models will be able to call patients and obtain variables about what happened since the last episode until the current one, asking questions of patients. These models will also be able to analyze all the blood samples and medical images of that patient and correlate these with the previous ones. These models can generate, for instance, an evolution of a CT scan in one year’s time and compare this with the previous one. When we talk about generative AI, this is one of its superpowers. So when the patient gets to the doctor’s office for their visit, there was a lot of work and preparation that was done up front, leaving much more time than there is currently for the doctor-patient relationship.

Then during the visit, we already discussed the doctor-patient conversation being captured by an AI agent. This is already happening in the United States with Kaiser Permanente, where they are deploying this at scale. But these AI agents can be used for inpatient visits as well. For a long stay in the hospital, there are a lot of areas in which these agents can help, running 24/7, from summarizing the clinical notes to looking at the lab results, the medication the patient is taking, and their vital signs. These agents can, for instance, identify drug-to-drug interactions that the doctor might have missed and signal to the doctor that a specific prescription may need to be changed. So, having these agents working for you, even at the end in helping prescribe medication, with a doctor always giving a thumbs up or thumbs down, the doctor is always in the middle, so you will never leave the AI running in the wild. 

We also need to think about aftercare. Consider, for instance, that the United States has 68 million chronic patients. When you think at this scale, you can imagine having models that call patients after care to check on the patient. For instance, they can see if the patient is checking his blood pressure, they can check if the patient is taking a new drug that he didn’t tell the doctor about, and then run through the entire internet to check if there’s an interaction for that specific drug that the doctor should be aware of. 

So, we will see this digital workforce working side by side with clinicians in ways that we’ve never seen before. 

Federico Guerreschi: JP, thank you for taking the time to speak with us today. 

Thanks for listening to The Heidrick & Struggles Leadership Podcast. To make sure you don’t miss more future-shaping ideas and conversations, please subscribe to our channel on the podcast app. And if you’re listening via LinkedIn, Twitter, or YouTube, why not share this with your connections? Until next time.

About the interviewer

Federico Guerreschi (fguerreschi@heidrick.com) is a principal in Heidrick & Struggles’ Milan office and a member of the global Technology & Services Practice.