Making AI Work Across the Supply Chain: Tecsys on Data Discipline, Execution, and Network Effects

Eighty percent of the data is outside of your four walls when it comes to the supply chain.

The first and foremost is getting your data house in order. That starts with a data audit, understanding what kind of data you have, understanding how you're using it, how you're storing it, how you're going to collect it in the future. That means a collaboration between supply chain, business, and IT.

Once you've got that in order, then we can start talking about some new AI tools to apply to it.

When it comes to AI and UAI within supply chain transformation, what I'm looking for is how we leverage the tool in different pieces of the supply chain. For example, let's take the warehouse.

When it comes to supply chain execution, I think there's some great use cases for, what I call, narrow AI, which some would call agentic AI, to look at things such as dynamic slotting, route optimization within the warehouse, so taking a dataset that is existent, that we know the data that we've leveraged, knowing the rules.

Also, knowing what the outcomes we're trying to achieve, and then applying an AI agent or a narrow AI use case to that. That's one area.

The other one, too, is things such as last mile optimization. We think about things or companies like DHL who are doing this, taking data from, let's say, Barcelona and understanding, obviously, the roads haven't changed that much. The houses haven't changed that much.

Taking that dataset and applying some type of AI to it to do a better job of optimizing last mile routing based on things such as traffic patterns, weather, etc., but understanding that the city itself hasn't changed much, so being able to take that dataset and applying AI to it to come out with a better outcome.

The advice I would give companies in today's era trying to leverage agentic AI, it's pretty simple. It's to get your data house in order.

From that perspective, it's understanding your data policy, data governance, data hygiene, how clean and good your data is, and also understanding the data sources. Because when we think about a supply chain, data's coming from a whole host of different areas.

By some accounts, I think Aberdeen Research said something like 80 percent of the data is outside of your four walls when it comes to the supply chain.

The first and foremost is getting your data house in order. That starts with a data audit, understanding what kind of data you have, understanding how you're using it, how you're storing it, how you're going to collect it in the future.

That means a collaboration between supply chain, business, and IT. Once you've got that in order, then we can start talking about some new AI tools to apply to it.

When it comes to data and life sciences, there's a whole host of areas that we can see some value in. We talk about things like patient outcomes, so simple things like triaging patients, understanding if a patient has an issue, understanding where the priority is for a patient when they come in, whether it's the ER or just the hospital in general.

Of course, we're seeing things like AI being able to do a better job diagnosing. Having said that, I would say the people still have to be involved in it. It's a tool.

Then we're going all the way down the supply chain, things like ordering, reordering, replenishing, looking at expiration, looking at things that are out of stock, looking at things that are going to expire or things that are going to perish within your pharmaceuticals.

There are a lot of very...back again to this notion of a narrow AI use case, where you have a tangible outcome, tangible rules, and a good clean dataset you can start looking at and then applying some AI tools to it.

There's a lot of opportunities across the health care/life science supply chain, everything from the patient all the way back to sourcing, manufacturing, procuring, and movement of goods.

There's a lot of opportunity, but the caution, I would say, is don't try to eat the whole elephant. You got to eat it bite-size at a time, which means, figure out along that supply chain what are the areas that you can apply AI to solve?

For anybody who's trying to get started on this journey, life sciences or otherwise, first and foremost, I'm going to sound like a broken record, get your data in order. Understand that at the end of the day, no matter what digital tools you apply, if the data's bad, it's the old notion, garbage in, garbage out. That's the first part to really get into this.

The second thing, especially for companies in the health care space, or any company for that matter, is to ask the hard questions. What I mean by that is when you come to vendors, supply chain vendors, technology vendors, etc., and I'm one of them, a lot of us will tell you a lot of great stories and maybe sometimes sell you a bit of magic beans.

For those in the industry, you have to ask us the hard questions. Why am I applying AI to this? Is it the right case? Do I have the right data? Do I have the right people within my own organization to leverage these tools? Are my people trained on this? Do I understand what the technology is? You're promising X, but do I really need Y?

That's the part that people need to take that discipline, is to be able to push back on those of us who are selling this technology, to push back on us to truly tell you exactly why you are using this tool and how you are going to get outcomes and then measuring that. What are the benefits I get from it?

Has the train left the station? To some degree, yes, but you know what? That caboose is still going past you, so you can still jump on it. Be prepared to ask those hard questions and to take a hard look at yourself. Do you have the right infrastructure in place, whether it's the data, whether it's the people, whether it's the ability for change management, to take advantage of these tools.

When it comes to new innovations we're seeing in the warehouse, so the execution side, I think it comes down to a very, I want to say, simple yet very powerful thing that companies like TraceLink and ourselves are embarking, which is, again, that notion of supply chain visibility, the ability to truly see what's happening at different nodes of the supply chain, the warehouse being one of them.

What I see coming down the road, which I think would be really exciting, is when we can tie in things outside the warehouse to understand how those actions impact what's happening in the warehouse and vice versa.

That can only happen with greater visibility, a greater digital platform that can allow you to tie in different sources of data to get a true picture of what's happening, and then the levers you can pull or push to get different outcomes. I think that's exciting.

It's not here today, but as we build towards it, I could see that in the future. The future being in the next couple of years, you would also be able to be much more proactive when it comes to your warehouse with regards to what's happening outside your warehouse.

When it comes to health care, a big trend, obviously, was outsourcing and relying on third parties to do a lot of the work that now, because of greater digitization and better knowledge of supply chain, if you will, we can see more hospitals bringing it back in.

That's a really interesting trend because what it's allowing these hospital networks to do is to get more control over their products, over how they can service their patients, over their relationships with other suppliers.

On some levels, we're seeing it's even allowing hospitals to create new business models where they can themselves become a supplier to other hospitals, to other health networks.

That's interesting, but it takes discipline around, again, back to what I've been saying about the data, understanding what data you have, having a platform that can allow you to see across multiple data sources. Then, also back to the people. You have to have the right mentality and the right people in place to execute on this.

I do think from the hospital side, this outsourcing side, we're seeing a trend where these IDNs are saying, "Hey, why can't we control this?" and, "Oh, by the way, now we have the tools to do so, so let's try and get this done because it helps us with our own patients, and it helps with our business overall."

When I look at MINT, I think the value prop that it brings to us and to everybody in the industry is the ability to do that one-stop shop in terms of getting access to data.

It's funny. I used to work for a company, GT Nexus, which had a lot of similarities to this, which is the value prop if you plug into this network, and now, whatever is already plugged in, you get access to.

When you think about that value prop, it just scales up from there because it's truly the network effect. Every extra node I add, the incremental value it brings goes up. Therefore, whoever's already on the network is going to benefit from that.

When I look at MINT, what's really exciting about it is that we're now able to connect once multiple times.

What that does is it encourages the next person or the next hospital or the next company, or whomever, to connect into that network because then they get the advantage of being part of that network. They get the advantage of the data that's already in there. Then those that are already in MINT get the advantage of adding new data sources.

There's this symbiotic relationship between new data sources, old data sources. That's the value of a platform such as MINT, which allows you to access a lot of different data sources, which if you did it individually, as we heard today from the main stage, would take you probably way too long than you had time for.

Where it comes to more value in terms of more digitization within the supply chain, I think it goes across the board. What I mean by that is if we think about the SCOR model, so plan, make, move, etc. At each stage of the SCOR model, if you had more digitization, you could see tremendous value. For example, we just talked about execution, but let's talk about things like planning.

If I look at planning, if I had greater visibility into things such as demand sensing, into historical data, into where capacity is within my supply chain, whether it's at the source level, whether it's the manufacturing level, I could be that much savvier in my planning because I would know what the capacity is.

Just because I could say, "Hey, I'm going to build 10,000 widgets," maybe my factory tells me, "Yeah. I can build it, but guess what? I only have raw material for 5,000 widgets." Rather than plan the 10,000 because one node told me they had the capacity, if I had a greater view across the supply chain, I could then be smarter about my planning.

That's one area where, I think, when you look at greater digitization, greater visibility, greater understanding of what's happening in my supply chain, it's going to hit all the nodes of the SCOR model and roll up into what we would hope is a more efficient supply chain.

At the end of the day, when it comes to the pharmaceutical supply chain, it truly is about lives. We're talking about people ingesting or consuming medication to make them better and make them healthier, cure disease. That's about life. This is not about a T-shirt getting there at the right time, whether it was a large or medium.

From that standpoint, we took this as a really important step in our relationship, our partnership with TraceLink because we saw the value that the partnership could bring to address the key notion of making people better in the health care supply chain. That's our North Star for both of us in this relationship.

Again, it's about patients. Patients are about health, and that means it's about people's well-being.