From Company Culture to Data Foundations: How Santen Is Preparing Its Organization for Agentic AI

What if the AI could help us to create a digital twin for the clinical trials based on the data that is already available so that you don't need to run preclinical phase one or two because you have the data already and AI can then validate based on the data that this drug is safe and is efficient and it can be manufactured with the GMP quality?

Then you go directly to phase three where you really have the large-scale human test. That would cut down the time dramatically and cut down the cost as well, making more and more better drugs fast available to the patients.

I think the most important thing is to help people to overcome the fears and expectations of working with GenAI and AI in general. You do that by motivating, inspiring them to be curious. What is this technology? How can it help me? How can it make me better in my job and how can it benefit the company at the end of the day?

Because you need to really incentivize them internally to find the right reasons that they would like to use them and then lower the barrier so that they can get started. Once you have that curiosity in place, then people start exploring, they start using, they start asking and they start sharing as well.

A kind of building bottom up, first people being curious, then being active, and then sharing, and at the end of the day coming together and sharing what they have learned and learning more from the others.

Then, of course, you need to have the supporting infrastructure in a company providing the tools, policies, and guidance and training as well. Have the people and the employees able to do that. First is really building this momentum and motivation and even movement inside the organization that people want to join.

Hikari is Santen's cross-functional, cross-regional initiative for bringing the AI capabilities for employees to use in their own work and in their own functions, but in a coordinated way so that we share the same tools, same platforms, even same processes as much as possible so that we can leverage the knowledge within the functions and between the functions and between the regions as well.

That we get this momentum and movement ongoing that everyone's really working with the same set of tools, policies, guidance and platforms as well. That way we get the scalability and leverage how the use of GenAI and AI in general can move forward.

We have 13 different projects in eight different functions cross-regionally, where we are trying to achieve, of course, obviously different objectives per function. The overall theme in all of these functions is how we can transform our current way of working, maybe also our processes, and especially our mindset.

How we work with this new technology it's going to help us, but how do we make it help us so that we can still maintain the trust and integrity that it produces and how we can apply that to our jobs and in our daily business?

Governance is essential. Governance brings structure, and we need structure in order to achieve the collaboration and knowledge sharing within the company. The essential thing for governance is executive sponsorship.

Without that, things won't happen because otherwise you don't get the budget, you don't get the resources, you don't even get approval to do it, and you don't get that as part of your milestones or your objectives in your work.

Once you have that type of support from your line manager or even from the executive team, then you can be sure that this is something that the organization wants to do and I'm part of the movement.

That hopefully then creates the motivation and curiosity that I really want to do this. Then, things start happening automatically. You don't need to push people anymore because they know that, "I want to do this, people expect me to do this, and I want to do this."

It's like having the same set of rules and same set of practices and systems and platforms so that we get the leverage, so that we avoid the siloing and we avoid the people just doing their own way. They can do things their own way, but when they have the same set of tools and processes and policies, they end up actually working very much similarly.

That will then help the organization to scale fast because people are speaking the same language and using the same tools. It's easier to share the experience. Additionally, what is really critical is that this way actually the knowledge stays in a company. If the people leave the company, usually the knowledge walks away at the same time.

Now, when they use AI, the knowledge stays in the company. You actually start accumulating this silent knowledge which is not there in SOPs, not there in documents, not there in processes. It's in your AI. AI can then tell the other employees that this is how we've done this before and this is how I can help you.

Actually you start gaining organizational knowledge and understanding there, which is supporting the other people regardless of whether the person who actually did that job or knew that thing before is not there anymore. I think analytics and agentic AI are two different things.

They support each other, but in agentic AI, it solves the biggest challenge at the moment, how do we manage the data so that it's useful for us? With analytics, we get the data. We, obviously, as employees and experts, we analyze and then we come up with the analysis of whatever results we want to get, but with agentic AI, we have a coach.

We have a trusted partner who will then support us in looking behind the actual numbers or data and maybe gathering information that we didn't even expect to be relevant for that particular analytics part. If we look at the challenges we have at the moment in order to get there, we have a multitude of systems where we have the data.

We even have analytics data, but they are siloed in the different systems. The challenge is, how do we get the agentic AI systems access to that data, to the silos? The systems are very much black boxes, and the data is within the walled garden. Without that, it won't happen.

Once we get the possibility to have access to our own data, because we own the data in those systems, then we can start accumulating and enriching the data with agentic AI. With the business knowledge and process knowledge it has, suddenly we can have even predictive types of capabilities where we can start looking at, not what happened, but what could happen in the future.

That's a completely different ballgame compared to the analytics which is giving us a present state of understanding of the business or operations we're having. MINT complements agentic AI because, like I mentioned, we have data in silos.

We also have data outside of our organization which we need to have accessible in order to make better decisions and be more productive with agentic AI. MINT solves that. Using MINT, we have the capability to combine the data sources. The data can be in a different format. That's not a problem. We don't need to have any type of data normalization.

AI can take care of it, but AI cannot function if it doesn't have access to the data. This is what MINT solves nicely and very elegantly. MINT itself, of course, it can use AI to further process and even preprocess the data for AI needs. I see that MINT's biggest strength is in breaking the silos and giving the access to the data we own or we can have for our disposal.

I think AI will be so fundamental in all aspects, not only business but also life. The impacts are countless. If you look at the pharmaceutical industry, I can name three different areas where it might have a really fundamental impact in how the pharmaceutical industry works and what kind of business the pharmaceutical industry does.

First of all, AI can help us to get to hyper-personalized medicines so that the medicines now they are prescribed in a fixed format, but in the future, AI might help us to prescribe drugs which are tailored for you. Not only that they are made for you, but they are also tailored based on your therapy progress.

In some stages you need a little bit more medicine, in some stages you maybe need less or you need something to help with that. It monitors all the time how you as a person are reacting to this therapy and what is needed. It's not like a one off and then you go.

It's like a continuous drug, if you will, using the medication and also gives better output with the therapy. The second one maybe is where a lot of money is being used in the pharmaceutical industry is the clinical trials. Obviously, AI is already being used for supporting the clinical trials, documentation and analyzing the output.

What if the AI could help us to create a digital twin for the clinical trials based on the data that is already available, so that you don't need to run preclinical phase one or two because you have the data already? AI can then validate based on the data that this drug is safe and is efficient and it can be manufactured with the GMP quality.

Then you go directly to phase three where you really have the large-scale human test. That would cut down the time dramatically, cut down the cost as well, making more and more better drugs faster available to the patients.

Maybe then the third is maybe quite obvious, is that AI could easily help to create virtual doctors where you can really safely and with high certainty ask whatever medical type of a question, and it's been validated. It has all the access to all medical records of your records securely.

It has all the access to the latest research and your prescriptions and so forth, so it's your kind of a doctor, buddy. It helps you, not only when you ask it, but it also proactively tells that, did you remember to do this and that? Did you take your medicine? Did you remember to eat breakfast? So forth.

It becomes integrated within your daily life. It helps people to be more aware of their health and more aware of what they can do to feel better and feel healthier.