Evolving the Supply Chain Mindset: How AI is Reshaping Behaviors, Roles, and Operating Models—Insights from Genpact

There is a cost of sharing data that we need to bring down. There is a cost of taking decision that we need to basically speed up. We are looking at it from a completely different perspective than we have looked at in the past. This is not going to simply cut resources in order to make efficiency.

What we're going to do is to bring the efficient to the next level by increasing collaboration, increasing the value of the information by adding on top of the information that's sitting outside the organization. Basically, you can create more value, defend yourself from threat, and catch opportunities.

Agentic AI is a huge opportunity for life science because it's going to finally change the way that the entire ecosystem is structured from a traditional manufacturing approach toward a more continually living system entity.

When you look at life science, when you look at the supply chain, we realize that we produce things in batch. We produce things according a sequence of task. We roll back if something goes wrong in a fashion that is much closer to production of goods. What agentic AI in general is going to bring in is a completely different way of managing the volatility that we are currently experiencing.

We're doing that by introducing three characteristics of living systems -- the ability to sense where the change is going to happen, the ability to adjust the change, and the ability to learn from it and reproduce basically the supply chain in order to move in a different direction. What is really special is that it is happening right now.

We are seeing a change that is impacting our current way of working that is creating a new way of reacting to changes. One clear example I can bring is a little bit our solution on clinical trials.

We have now the opportunity to sense what is happening in, for instance, the consumption of kits in a hospital, immediately understand where we could supply that for reducing that impact, and, in a way, protect the patients by making sure that they receive what they need in order to continue trials.

This type of approach of sensing, adjusting as well avoiding because that information remains in the loop. In the next planning cycle, we can be more effective in understanding the impact of variability is a real example that we can immediately have a very strong impact on the life of patients that are in the trials.

As well, the execution of the trial itself, which is one of the greatest improvement that we can make currently in life science. Speeding up the clinical-trials process is something that is immediately bringing value to the organization. In the end, it's bringing value to the patients because the patients are going to benefit from it with early treatments available in the cabinet.

One of the reason why we partner with TraceLink is the immediate availability of a network that can facilitate the transfer of data, value, and intelligence into the overall ecosystem. I'm going to use a sort of analogy to explain that. When we look at MINT, we're talking about the soil.

Let's look back to the living system organism that we're talking about to better understand the contribution of AI, agentic AI, in the current world. As a living system, what we have is MINT as a soil. Basically, it's the fabric, the connectivity, that's going to bring the vital elements to make a plant flourish. Outpost represents the metabolism of the plant.

Basically, each single organization is going to configure itself in the way that wants to function. It's the way they really function it. [indecipherable] are the tools that are used in order to grow and make that specific living organism prosper. Where are the humans and the AI into there? They are the gardeners.

They are the ones that are going to basically use the tool to adjust the metabolism and best use the vital elements in order to make the plant flourish and grow. That's the mission that I think TraceLink is accomplishing pretty well at the moment. The beauty of our partnership that we have a complementary mission.

While TraceLink has the scope of commoditizing this sharing of data, value, and intelligence, we have the scope of moving organization from a process-driven approach to a behavioral-driven approach. What we need to do is to transform the way that organization transform themselves. To do that, we need to look at the behavior. We need to basically leverage the information that are coming.

Not to basically achieve outcomes, but to learn what to do with that in adjusting our self into the future. We're talking about fragmentation, friction, and failure as the major costs that we are currently having and experiencing in our life-science ecosystem. When you're thinking about the cost in general, it is not any longer in the material. It's not any longer in the transportation.

There is a cost of sharing data that we need to bring down. There is a cost of taking decision that we need to basically speed up. We are looking at it from a completely different perspective than we have looked at in the past. This is not going to simply cut resources in order to make efficiency.

What we're going to do is to bring the efficient to the next level by increasing collaboration, increasing the value of the information by adding on top of the information that's sitting outside the organization so that basically you can create more value, defend yourself from threat, and catch opportunities.

This is a little bit one key difference when you look at AI compared to other automation tools. AI is offering you the opportunity to take care of the roots, which is not a simple simplification of what you're doing -- a cover-up. You can go really into the detail. By doing that, you transform your way of operating.

That's why it's really important that that goes together with a behavioral analysis of what you want to do. We need less stopwatch and more compasses in the future. In the past, I remember when I started in automotive. One of things that they were doing was measuring the time.

It was called work analysis -- measuring the time that the worker was spending on the supply line and trying to understand how that approach was going to simplify. That's not the way it's going to work any longer because we have changed the approach. We do not release any longer functionalities. We are releasing model.

We need to understand the use of the human intellect to understand how that model can basically answer eliciting needs that are not scoped in a simple system. We're talking about the way that organization are changing. We need to understand that, if we want to change a behavior, probably we need to change as well the semantic -- the names of the roles and the profiles we are looking for.

At the moment, everything is universal. We have managers. We have directors. That tells you already a lot about it, right? We are managing things. We pretend, in a way, to measure things comparing to the history and to the past. In a high-speeding world, that's not longer possible. Probably instead of manager, we should use evolvers as a role.

Somebody that's able to evolve the model is able to evolve the functionalities into something completely different by learning how to better use. The measurement is not going to be more of the KPI in the traditional sense. Probably the only KPI that will remain will be the ability to learn. Instead of direction that gives you a very clear idea, probably we need to go to our governors.

At the end, if you look at when you have a forest, you don't use the word "managing" the forest. You don't use the word "cultivate" forest. You govern the forest because it's an entity that has its own way of adjusting and adapting. You need to play with it. That's a bit of an analogy that I'd like to use to explain that.

Organization leaders need to change the way they approach performances in general. I would start by saying they need to change the question they're asking themselves. Until today, until now, the question was, "How can I simplify this? How can I reduce the cost there? What can I release to the machine to do it on my behalf?

" What do I need to teach the machine so that, basically, I don't have to take care of?" This isn't behavioral change. Again, we need to move from process organization to behavioral organization. We need to keep this continuous of sensing, adapting, reproducing as the key elements of it. I'm going to give you an example of where AI can completely change the way we perform.

It's moving out the organization from a static approach toward a dynamic approach. Let's think about, for instance, the aerospace and the aviation system, where basically each time that there is a failure, there is a new playbook. There is a new protocol. That is shared all around so that everybody can learn from that failure and avoid that the failure happen again. That's a static approach.

Well, working with AI is something completely different because each time that there is a failure, we're going to have an agent created to manage. That agent would become an antibody of your network that is going to help to protect yourself from there. Each failure we characterize a different antibody.

All of them together will grant the ability of the network of self-healing in itself. This is the approach that we're looking that we're going to experience. It's going to shift completely the way we are going to manage things from static to dynamic.