Confessions Of Supply Chain Executives | The Agentic AI Wake Up Call

Here's today's uncomfortable truth.

Or maybe not so much an uncomfortable truth, but a question.

If you're a retailer, should you have a clear plan for agentic AI in your commerce operations?

If you don't, are you at risk of already being behind?

I don't know.

But everyone sure as heck at NRF was positioning things with that degree of melodrama.

So I went to the experts for today's interview, one of whom is the single man I trust as much as anyone, who to set me right on those questions.

And here's the confession you're going to hear from both of my guests today.

Most retailers are completely unprepared for agentic AI.

They don't understand what it is.

They don't know how to evaluate it.

They don't have the data infrastructure to support it.

And they're making the same mistakes they made with previous technology waves, waiting too long, picking the wrong partners, or worse, rushing into pilots without a real strategy.

But here's the good news.

It's not too late if.

If big if.

If you know what to look for.

So today we're going to give you the roadmap.

A real one for what it takes to succeed in agentic AI commerce.

Welcome to Confessions of Supply Chain Executives, the podcast where we get brutally honest about the challenges, failures and celebrate the victories in retail supply chains.

I'm your host, Chris Walton, and today's episode is different.

This isn't a product pitch.

It's just three people who love retail and two of whom know far more about the nuts and bolts of making an omnichannel supply chain work at scale than I ever will, sitting around discussing the realities of agentic commerce.

My guests today are Omar Akhila, the SVP of product at Infios, and Adil Kazmi, the head of AI product development also at Infios.

They're working with over 5,000 customers across 70 countries and.

And they're seeing firsthand which retailers are getting AI right and which ones are setting themselves up for failure.

We're going to talk about what agentic AI actually means for commerce.

How to know if you're ready for it, what infrastructure you need in place, and most importantly, what mistakes to avoid.

Gentlemen, it is with great pleasure that I welcome you both to Confessions of Supply Chain Executives.

Pleasure to be here.

Glad to be here, Chris.

Thank you.

Yeah, so, I mean, Omar, Omar, you and I.

You and I, I was telling a deal and everybody, before we hit record here, you and I go way back like you, 15 years you've managed to stay friends with me for like 15 years.

So it's been hard.

It's a hard thing to do to say your friends for this long.

But yes, no, I'm honored, sir.

Absolutely.

And thank you for having me.

Yeah, for sure.

Yeah.

Well, like I said at the outset, and it's really true, like, you know, you, you are my go to guy whenever I have a question about anything like this.

And bringing a deal, a deal into the conversation too is only going to augment the knowledge that you can provide to me as well as to our audience.

So, so let's start with the fundamentals.

Let's get the fundamentals out of the.

Because I think when most retail executives hear a genetic AI in air quotes, either one of two things happens.

Either they glaze over and they assume that it's just the next buzzword, you know, after machine learning that they had to adjust to for a while, or this can happen too.

They get so excited they can't contain themselves.

So Adil, you're the, you're the AI expert.

So explain agentic AI in terms that a retail executive would understand.

What is it and why should they care?

I mean, before we even get to gen 2 ki, let's maybe take a step backwards and just define what is AI generally or generative AI in the modern context.

AI is not new, right?

We've had nlp, OCR and multiple technologies available to us for many decades.

The key unlock a few years ago was that up until that point, we machines could not understand unstructured data.

So let's say a conversation between you and I in a text format.

You know, a good analogy is call centers have been automated for many, many years.

I remember a couple, you know, five, 10 years ago, I'd call in to my local retailer with a, with an issue and there would be a robot on the other end.

But that robot was not able to understand semantically what I'm looking to say.

So if you recall a few years ago, that robot would, you know, say press one or press two or it would only be able to interpret it very basic syntax like yes or no or thank you, today's call center agents as an example.

I mean, I can speak to it the way that I'm speaking to you, Chris, and it will totally understand what the intent is.

And then based on that intent, it can reason, hey, what is Adil looking for?

And then once it's identified the reasoning, it can then finally act.

So if we use that similar framework, gen AI is the ability for machines to understand unstructured text, which is a key unlock.

I mean, for millennia we have not been able to program machines to be able to understand this.

So that was a key unlock.

And then agentic AI is the ability to augment gen AI with actions.

And so, you know, a good framework to thinking about what is a gentech AI is it is the ability for machines or generative AI models to sense and listen, reason on their own.

And there's multiple frameworks, there's looping mechanisms, et cetera, et cetera, there's react frameworks.

And then finally, and most importantly, what makes it agentic is action and whether that action is helping a customer process a refund or letting the customer know when their expected delivery is, et cetera, et cetera, et cetera.

So agentic AI is the combination of listening, sensing, reasoning and then most importantly, acting.

Got it.

That makes sense.

So it's basically like the difference between recommendations and taking action.

It's the ability to make decisions on unstructured data too, which is the other key point that you brought up.

So when we start talking about commerce, then a deal like, you know, and we, and again we talk about agentic AI and its impact on commerce, do you think it's going to have the biggest impact on the customer facing applications in terms of what you just described, or will it be on the back end operations or both?

And which comes first?

How do you think about that?

It's impacting everything, the entire value chain.

That's going to be everything.

I'll share some quick anecdotes.

We were at NRF just a few months ago and there was an announcement of a ucp, a unified commerce protocol which enables retailers to actually, in a single format, propagate their products to users in ChatGPT.

So that's, you know, touching.

That's agentic AI touching the customers and the retailers journeys.

Right?

How retailers meet customers, where they are and how customers discover products.

Last I checked, you know, I think ChatGPT has over 800 million users, active users.

That's a massive, massive, you know, channel, which is which of which we're only.

Retailers are only touching the surface of, you know, we took the example of customer support operations.

That's another area where Gentek AI is already being applied.

Forecasting, you know, predicting very intelligently what expected delivery times will be and showing that at the checkout.

So agentic AI, it's hard to say that there's only one place the entire journey is being redefined in real time.

Right.

So Omar, Omar, I want you to expand upon that then, because I know from talking to you over the years, as I have, you've worked with a lot of different retailers and brands.

If you expand on what he just said, where are you seeing people run to or try to implement agentic AI where it provides the most value for first?

Like so, you know, if I said that another way, are there quick wins versus, you know, longer plays as, as people are grappling with this, you know,

piggybacking on what ADU just, just talked about.

If you think of, you know, the old days and, and you know, from IVR technologies where people would basically.

And IVR is what Adida was talking about, about the call center technologies.

It was innovative, you know, integrated voice recognition technology.

So where, where you'd call and basically you, you, you'd get a robot.

We deal with it with Wells Fargo, with Apple, with all.

So when you think about augmenting your staff today to handle calls like, where's my order?

But now taking it from just a receiver to actually a system of action to actually help you do things like placing orders, changing orders, right?

We always were able to do things up to a point, right?

So we can say, oh well, we could provide order status, but we can't do order modifications.

That's, we don't have the integration in.

That's too hard to do, right?

We can't do, you know, book a new order, right?

So when, when you start thinking about supply chain execution and all the areas and all the amount of, of of, you know, support that you need to do menial tasks, like somebody calling and saying, okay, hey, you know, where's, you know, whether it's on the order side or even on the transport side, where's my stuff?

Right?

Can I book a new order?

You know, this ETA that's coming in, right, this, this product that's coming in with, with an eta, how does that going to impact, right?

My present order, there's a lot of areas, you know, to be able to plug in the agentic side of what, what ADIT is talking about, right?

And I think that, you know, we're, we're very right and I think retailers across the board need to be very purposeful about the problems they have.

And you've heard me say this for time, right?

So a lot of times we look through the technology lens as opposed to the strategy lens.

Where do I need to go?

Where do I have a problem?

Do I have a problem, you know, with, with, with labor?

Do I have a problem with, with analyzing the data that I have in front of me and really apply because to add this point, it's so available now, you need to really understand what your strategic unlocks are for the company and then apply it to that area.

And we believe that the execution, again, a lot of it's happening right now in the planning world as well.

But we believe, but I think that's more of the predictive side when you talk about agentic.

I think execution is, is, is absolutely, you know, ripe for it across, you know, you mentioned it, whether it's agentic commerce, right?

All the way through to helping optimize, making sure you're making the right decisions to tracking orders, to booking orders, to changing labor.

What should I do?

Something, you know, 10 people called in sick.

What do I do with labor?

There's all kinds of great use cases.

But again, to me, you need to anchor in the strategy of where the company needs to go and use it as an unlock for that strategy.

That's a great point, Omar.

Because, you know, it's funny, like, as you're saying that, I'm thinking back to all the conversations I've had recently about hearing that, you know, the boards are even saying, like, what's your AI or agentic AI strategy?

But that's the wrong question.

The right question should be what is the strategic goal that you want to achieve and how do you deploy AI or agentic AI to help you solve that?

How do you use AI?

And I, and, and again, I'm going to steal what I've heard AD say and I'm sorry, Aden, I'm going to, I'm going to steal something.

You said too often in the, in, in the world today, folks are looking at technology to solve, right?

And they're saying, okay, well how do we use it to.

No, what this is, it's an enabler for you.

At the end of the day, your strategy is still your strategy.

You now have more robust tools to throw at that strategy, to throw at that.

And you should evolve your strategy, obviously, because of where the world is today.

But I feel like, again, to your point, people are looking at it to say, okay, there's this cool shiny box of agentic AI.

What are all the things we can do with it?

As opposed to let me look at my strategic needs across my company that I've not been able to solve.

Whether it's been, you know, volatile labor, you know, it's been, you know, managing my carriers and ensuring that I know where they are, new bookings to, you know, etc.

Look at your supply chain and say, okay, well, how can, can these new technologies that are here and even though I was saying they're not new, their application is new.

Right.

That we should now look at and say now I have a big unlock with this, you know, concept of contextual.

Right.

How do I apply it?

Where, where, where can I apply it to unlock my, my, my, my strategy?

Yeah, I'm having beacon nightmares as you're talking.

Like remember beacons when beacons were like rage for like 15 or 15 years ago.

It's the same issue, you know, but you know, this is a little bit different scale and scope.

But, but I'm curious, like when I hear Omar talking, I hear what you say though.

Like the one thing that goes, that comes to mind for me is like I think if say my strategic goal, and I think everyone has this strategic goal which is to run their supply chain more efficiently, more profitably, have their products in stock more often, that feels like a place ripe for agentic AI to me that it's going to hit that area first because not only is there a need there, but going back to your point Adil, the data is generally speaking more structured and the ROI is more measurable in supply chain operations and it's, than it is at say other parts of the business.

But do you agree with that or am I thinking about it wrong?

Is there nuances to that question, Chris?

I mean I, I 100 agree and just.

You do?

I, I do.

And, and just taking, you know, exactly what Omar mentioned, he mentioned something really interesting which is two things.

One is that agent AI being a force multiplier and that execution or supply chain execution is the prime target or the perfect fit for agentic AI.

And the reason for that is, I mean, so let's, let's maybe and take a step back and look at this problem from first principles.

The first is that supply chain execution has never been more volatile.

You know, everything from macro things that we do not control, tariffs, global supply chains being disrupted, you know, the ports closing down because of XYZ reason or another, and then even micro, you know, Omar mentioned late quoted labor volatility being an issue, stock outs.

Right.

Where's my, where, where, where physically are the goods?

Where are my customers?

Execution for supply chain leaders has never been more difficult to operate.

And agentic AI is a perfect use case from both an ROI and an application perspective for the following reasons.

Number one, humans can only process so much data, right?

The data.

I would argue the data is already there.

Most retailers have made the investments into the systems and to the applications to capture the data.

But now the next leg is how do you action on the data that is available to you in a real time way.

So it's one thing to say that hey, I've got the data and I'm looking at it three hours after an event.

Wouldn't it be nicer to be able to have something that's autonomous, always on listening, sensing to changes in your operation and then based on your processes taking action.

And that's really the promise of agentic AI and supply chain execution.

Let me share an example with you and this is an example that you know our customers, we're working with our, with, with our own customers on.

You have a retailer, a customer, a shopper like myself comes on today's Friday and you know they come.

A shopper like myself arrives on the retailer's checkout and I see an EDD of Wednesday.

So the shop, the retailer is now promising that if I check out right now, I'm going to get this delivery on Wednesday.

But we all know that we live in a really volatile world where supply chain is now being disrupted daily.

So in the old world I would check out that order, that promise would then go into my 3 PL warehousing things would happen inside the 4 walls and inside the warehouse.

Then the physical good would be handed off to our logistics partner or carrier partner who would then be responsible for doing the final mile delivery to my doorstep.

Supply chain execution and operations is messy as you pointed out Chris, because while the data is there, supply chains run based on intermediaries.

Very few retailers are vertically integrated from promise to final delivery to the customer.

There are multiple people, multiple companies that are involved in moving that good and taking the order, et cetera, et cetera, et cetera.

So in with agentic AI, what retailers can apply as you know, real use case today is you could have a real time visibility across your entire order life cycle.

So once that order is picked from the warehouse, once that order is put on the truck and it's on route maybe in the middle mile, let's say that the driver, the freight driver is unable to make their, you know, their next destination to New York City on Monday.

Now you can have an AI agent that's listening, acting and sorry sensing and acting it sees that this driver will not make the delivery to New York City's warehouse on Monday which then downstream will affect the promised delivery date to me, the shopper for Wednesday.

Now the AI agent can see that anomaly, detect it, reason, act and acting in this scenario would be, could be dispatching an email to me, the shopper and saying, hey, we need to reschedule you to a Thursday delivery.

Or if we think even upstream in the warehouse, if the pickup is not on time, then autonomously, agents can reroute that order or pull it from a different warehouse.

These are all examples of how retailers can apply Gentech AI throughout the order lifecycle process to begin operating more autonomously.

With the key goal going back to what Omar was saying.

Your strategic objective is to keep your customers happy and keep them coming back.

So LTVs are increasing and the customer promise is not failing.

This is an example of how we can apply gentech AI in our supply chain operations to meet the promises and the commitments that we made to our customers.

Yeah, that's a great, that's a great example, you know, and it calls to mind to me something that I've, that I've learned very early on in my career at the Gap, which is like to your point, there's always inaccuracies along every step of that process too, where human intervention has to come into play or some type of decision making has to come into play.

The example I always remember, which is very similar but in a different context to the delivery driver, is like, you know, somebody comes in and swipes all the sweaters off my table and suddenly the system thinks I have 12 sweaters sitting on that table.

But I don't.

But, you know, through AI, you could understand, oh, I should be seeing sales in those, but I'm not.

And I could trigger some type of action again to go and tell the store, look and tell me if those sweaters are there and if they're not, then let's take this action and get you more sweaters.

And so those types of things, like whether it's the driver, you know, getting sick and missing, you know, not being able to deliver the product, or somebody stealing a batch of sweaters that makes supply chain complicated even beyond this more standardized data that it has.

So if we buy into that, like what kind of payback are we talking about financially?

Like to make in roads on this to get the value and benefits of the gentic AI ideal.

What type of payback period are we talking about here?

Is this still going to take years to see the fruits of this or can we shrink that down?

Is the value capture such now that we can show benefits from this much faster than ever before?

I'll share live anecdotes with customers that we work with.

And again, Omar mentioned it perfectly, which is being purposeful, not just throwing technology at the problem for the sake of, but aligning your technology investments with your strategic objectives.

So to answer your question directly, what we're seeing is that based on the workflow and the use case, the ROI is near immediate.

Our customers really approach agentic AI for one of the following reasons.

Number one, they're looking to do more with the same.

And an example of that is, if we think about, you know, retail is a great example actually for this.

Your demand curve is very cyclical.

In Q4 you have peak demand because of, you know, we've got the holidays around us with people are buying.

And then in Q2 demand kind of, you know, cycles down for a bit.

But your labor, the people who are responsible for moving your orders through the life cycle is very much so a fixed, a fixed cost and it's a capacity, right?

So in Q4 you have peak demand up here and your capacity to serve is very much so fixed because labor, you, is, is treated as, is a constant in many ways.

So you have that gap, right?

Demand is up here and capacity is over here.

That gap is your customer experience.

If you're not able to provide customers with that always on consistent experience, they may not come back to you.

In Q2 when demand falls off, your capacity is over here.

So now you've got capacity here, you've got demand down over here and that's wastage.

So you got more than what you're actually consuming.

So the first way to think about ROI is to smooth out your cost curve or align your cost curve to your demand curve.

As a retail operator, you do not have control over demand.

I mean, in many ways you do have advertising, etc.

But demand is very much so in a way out of your control.

However, your capacity is 100% in your control.

So reason number one, why customers are moving to Gentek AI is to be able to do more with the same.

And what that translates to, you know, to your CFO is being able to align your cost curve to your demand curve.

The second reason is actually exploring and doing things that you were never able to do before.

An example of that is, you know, we have customers who have said to us that we currently have visibility across 30% of our orders.

And when asked why, the answer was we don't have the tools, the people, the processes.

We'd love to be able to cover 100% of our orders in terms of visibility, but we can't get there in a cost efficient way.

So the reason number two is customers are looking at agentic AI as a way to improve their operations.

To be able to do things that they were never able to do before.

And I keep going back to visibility as an example, but there's many, many examples throughout that order lifecycle journey.

And then the third is really elevating people.

So Omar mentioned a lot of the tasks in supply chain execution are manual.

They're repetitive, they're tedious.

Humans want to.

Teams want to manage exceptions.

They want to graduate to being able to make decisions.

And that's again, where agentic AI helps.

Our agentic AI thesis is not to replace humans.

Rather, it's to create assistance to humans so humans can elevate themselves and work on the things that matter the most.

You're right.

Nobody wants to be a button pusher.

You know, they want to feel like they're making a difference in the decisions they're making.

And so that's a really interesting way to think about this too.

All right, I want to shift gears a little bit.

We've, we've kind of talked about the theoretical here, you know, so far, to start us out in the first, you know, 10 minutes of this conversation.

But I want to get more to the real now because, you know, one of the things I know from being in the industry for 30 years and from talking to companies like I do every single week, you know, there's.

There's somewhat a feeling, and I don't know what.

I don't know if I want to put a label on it, but, you know, I feel like most retailers dramatically, and I will use the word dramatically, overestimate their readiness to apply new technologies in their business.

They think they're.

They're like, many of them think they're ready for AI.

They want to go guns blaring into it.

But, you know, a lot of times they're still struggling with basic data hygiene.

So, Omar, I want to do a readiness assessment with you.

If there's a retailer listening, what are the prerequisites that you have to have in place before you can even start a conversation inside your organization about implementing agentic AI?

You know what's pretty cool is, Chris, if you and I would have had this conversation maybe a couple years ago, I would have said understanding the data that you need, et cetera.

But now to add to this point, because now with agentic and contextual, it's really about understanding the problem you're trying to solve, right?

So now it's actually going.

And I think this is so cool that this is where we are as somebody that's been in retail and in B2B for years.

We're now at a point where I don't need to think about the technology I need to line up.

I now need to think about the problems I need to solve and this, you know, and be very crystal clear that they'll yield the most benefit.

And I'm going to go back a little bit to what you and Adit were riffing on.

Yeah.

You remember L1, L2, L3 support?

You've got like, you know, hundreds of people that you're trying to screen first calls and second calls and third calls.

And then you need more labor.

Right.

I need a different type of labor source to add this point.

Now you have assist where you can reposition your best people to doing the most meaningful work.

And now you can deploy agents, scale them up and down without having.

So when you talk about roi, I no longer have to have a complicated, you know, model to support my peaks and valleys that I have in retail.

I can deploy agents at will.

I can deploy agents for specific problems at will.

When you think about, you know, and I'll talk about it a little bit later, but like, you know, where it's, it's, it's going is you'll be able to, to stand up and, you know, create an agent within a matter of minutes, not, not months, not weeks.

You'll be able to just say, here's the problem.

Go solve it.

Right.

Whereas, I mean, think of the staffing.

Think of the readiness.

Think of the things you have to do, Chris, in stores and in, in call centers and everywhere else to support a new initiative, whatever it may be, whether it's seasonality or peak or whatever else, that world is gone.

So now the, the, the, the real kind of solve is I just need to be clear on the problems I'm trying to solve.

And that's what I should spend the most time thinking about is where.

Where am I hurting the most?

And what would I like it to go and solve for me?

And that's a different conversation than we've ever had.

Right.

This is a conversation that we would have loved to have a couple years ago.

Well, yeah, I mean, yeah, I'm actually having trouble getting my head around this too.

That's why I love talking to both of you.

So you're flipping the script on me.

You're basically saying this is like the greatest gift that we could imagine from a technology standpoint inside of a retail operation.

Because if I hear you right, then I don't have to adapt my stack that much.

Right.

Like, this can just composably fit alongside whatever I'm currently running and actually make it better if I, But I have to be smart about this strategy of where I want to deploy it first and foremost.

Yes.

And let me just, and I know Adil will chime in in the past, right, And I'm going to geek out with you a little bit.

I needed to make sure that everybody had the same definition of an element across, right.

So order meant this in, in wms, this in tms this and that doesn't, that's not needed anymore.

With a gentic AI, you just need to set the context to say these two systems are talking to each other.

So an order means this when it, when, when they're talking to each other.

That's where, when you start going into.

So again there will be a lot of people that will be like, well, but you need, you still need the data.

Yes.

If you have a common data like it's, it's even better.

But at the end of the day, if you have a data of orders and a data of, you know, the shipments that match those orders, now it's, you know, just like you were doing in Excel with these drawings and analysts were trying to figure it out.

You can actually have an agent just do that for you and now figure out all kinds of analytics about it, right.

And ask a bunch of questions about ETAs.

And then to add this point, go from those questions to actual actions that, that you want to perform.

That is, is, is, you know, again, you'll get challenges, I'll be challenged quite a bit.

But I think what you are seeing in the world is, is this, it's not about needing to set up this tremendous infrastructure to support within the company.

Now your Googles, your Amazons, everybody's doing that for you, right?

That's, that's, that's where they're gonna.

But in terms of you as a company and adopting it, right, it's really more about making sure that you're clear on the problems you're trying to solve and having the right providers to help you do it.

I mean, I couldn't agree with you more, Omar.

The only two things I would add then, one, exactly what you said, like we don't actually need a single data lake, a single data model anymore to work with agentic AI from a readiness perspective.

Exactly what Omar said.

You modern LLMs are so capable.

And look, here's another thing.

They're only going to get better.

What gets released in 2021, next year and the year after will be even more powerful than what we have Today.

And I can tell you that what we have today can make sense of, you know, what order status means in this table and this table and that table and this table.

Exactly what Omar is saying.

I think from a readiness perspective, it's actually more so internal, it's more so culture change management.

And I'm going to piggyback.

Exactly.

Piggyback off exactly what Omar said.

What problems are we solving?

What is the process to solve these problems?

How do we leverage AI in the right parts of this process?

I think those are the biggest impediments to be actually adopting AI.

Because I would make the argument exactly as Omar did, that the models are so good, I'm going to say, you know, verbatim, if you have a data lake, fantastic.

If you don't have a data lake, you don't need it.

You can still adopt gen AI, agentic AI.

Both controversial, Chris, because everybody's going to be like, I need data lake.

These guys don't know what they're talking about.

The reality is you don't.

And you just need the right context.

And that's the power of this technology.

Right?

But.

And there's always a but, right?

There's always a but.

But this, That's a really good point that I think is going to be, you know, very interesting for most of the listeners.

But to Dill's point, you've got to have great talent and you've got to have a great organizational structure around this that knows how to strategically operate in this way.

Right?

And so that's my question for you.

Adil is like, you know, what does this mean from a talent standpoint?

Like, do I need to hire more data scientists?

Can my existing teams generally handle this?

Do I need to bring in vendor support?

Like what?

Let's get into that, like organization.

What are, how do you answer that question?

I'll share another spicy take.

You know, I come from, I come from startups.

I've spent most of my career actually building startups, being inside startups and like let's kind of, you know, align on what's happening today.

Startups are, everyone has access to the same models.

Everyone.

Whether you're Google or you're, you know, a two person startup that was just founded in dad's garage yesterday.

We all have access to the exact same models.

But for enterprises to adopt AI, they need way more than just, you know, a wrapper around a model.

They need workflow, redefinition support, implementation training, you know, model guidance.

How do we fine tune these models to the unique processes and to the, to the specifics of your business.

And so I would say that, you know, from the perspective of a retailer, you really have three options.

Option number one is that you do exactly as you mentioned, Chris.

You do invest internally to stand up the team to build that, to build that muscle memory that can take you upwards of 12 months realistically, and it's costly.

That's option one.

You stand it up internally.

Option number two is that you buy off the Shelf AI Agentic AI tools which will deliver immediate ROI, but only at the surface level.

The moment you want to start thinking about that entire, you know, that retail order journey and the examples we just shared with you, where one agent is responsible for looking at the transport journey, another is looking always on looking at the fulfillment journey, another one is looking at the promise and fulfill journey.

Off the shelf solutions were not customized.

They're very horizontal, they're very generic and by nature they're easy, you know, quick deployment cycles, but they won't let you autonomously operate your supply chain.

And then the third option is really where you partner with the right vendor that not only brings the agentic AI tooling, but partners with you and in many ways acts as an internal team member to do two things.

The first is helping you with the strategy, which is what problems are we solving?

How are we going to solve it?

What is the roi?

And then second, which is implementation.

And implementation consists of the following.

One is looking at the workflow.

You know, how did we do this job yesterday?

How do we want to do this job tomorrow?

In a world where agents and humans are working alongside with one another.

Evaluations, training the models, making these models specific to your operation, to your needs so that they deliver a higher roi.

And third, connecting.

So having agents, it's great to have individual agents, you know, touching your, your order execution journey at the surface level.

But it's amazing to have your agents not only automate work vertically but then horizontally speak to each other so that you can actually run your supply chains intelligently.

So those are the three models, the three, you know, from a retailer's perspective, do it yourself by hiring a team could take you 12 months or more and very expensive.

Two is purchase off the off the shelf solutions which only allow you to attack the surface level of problems.

Or three, select the right vendor that not only brings that gentic tooling but, but brings a deep domain expertise to help you implement it, redesign your workflows and then get to better outcomes.

My fear with option two is I feel like, and given the fact that you guys said you don't need a data lake anymore too.

I feel like you end up in, in pilot purgatory if you're not careful with that approach as well.

When we talk about the transformation of AI in, in, in the strategy, it's also transforming the team that supports it, right?

So you know, to add this point, the type of skills that you need, right, are actually in your product people and your people that actually understand your business operations, etc.

But now to get them to a point where they can actually, you know, have, you know, a gentic outcome, right?

To have this point, it's going to take close to 11 to 12 months to upskill your team to get you there.

So you know, the key is we're not just seeing the transformation across the technology, we're seeing the transformation also in how organizations need to realign to support it.

So what ad and mentioned in terms of how you support is absolutely essential, right?

To look at it very closely and say what is the fastest way for me to get there?

And you know, in most cases it's option, it's option three.

Okay, so let's put that to the test then.

So Adil, again back to you.

Like, if I'm going to take that third approach, what are the questions I should be asking to determine that if the partner that I'm choosing actually has true agentic AI understanding and capabilities versus just trying to spin up some good marketing?

That sounds good.

The first would be domain expertise.

Does this vendor understand the industry that I operate in?

What is their longevity here?

And that's why incumbent SaaS has an incredible opportunity to adopt, modernize and in many ways take advantage of Gentek AI.

And the reasons for that are the following.

SaaS, your SaaS provider, SaaS providers in your industry, they have access to contextual data.

They have teams that have deep domain experience, they can speak your language, right?

When you say you know fifo, when you talk about your warehousing operations, they can understand what you mean.

Omar alluded to it many, many times, which is what is our strategy?

What problem are we looking to solve?

You can reduce the time to value by working with vendors that have expertise in your domain.

And that's why I go back to vertical.

AI is, you know, the key opportunity from a vendor's perspective.

That's why we focus, we hyper focus on supply chain execution, the promise fulfill and transport order life cycles.

So Omar, then like when we go into that type of relationship then like what should, what should retailers expect in terms of partnership, ongoing support like you know, because if you're going to go down that third model, you know, there's a lot of, there's a lot of puts and takes that come with that model too in terms of like wanting to make sure you're partnering with people that are going to stick around for a while.

So like, so what should the retailers expect on that side of things and how do you structure those relationships financially?

Because you've worked on both sides.

You've worked for the retail, you've worked for the technology solutions provider.

How do you think about that?

That's a loaded question, Chris.

I think it's, it's two, it's two sides at this point.

You need a vendor that, that is very clear and has deep domain expertise in the areas you're trying to solve.

Right.

So that's the first is I'm looking at somebody that, that, that again, especially in today's landscape with what's happening, you're getting a lot of folks popping up and they're solving one particular thing, but they're not necessarily seeing the intersection and the connection between the function.

So you need someone that understands the domain well and understands that it's not just about one particular agent that can do one particular thing.

It's about something that's looking at the area.

Right.

As we talk about supply chain execution from orders all the way through to delivery and actually has a perspective on, on helping.

Right.

Firm out and complete your strategy.

Right.

That's one.

To your point.

I don't, you know, again, as somebody coming from the software industry.

Right.

And, and from a retail background, you know, I also need to be cognizant of the fact that, you know, the landscapes do change.

So at the same time I pair them with some of my best within my company as a retailer so that they understand how and what.

Right.

Like, because again I, because to add this point on the, the three options, I do fundamentally believe that you're going to see an upskill of the, of, of the workforce.

You hear people and we see them on LinkedIn and everywhere else that people are taking courses, they're trying to figure all this out.

So you know, I would look at it as an acceleration path, right, where, where you're partnering with, with, with vendors that have deep expertise, that have the longevity.

But at the same time you're also upskilling your team in the time.

Right.

That 12 month ramp that Andrew was talking about, you're upskilling so that they're working hand in hand.

Whereas best case scenario is, you know, you're continuing along the path together.

Worst case scenario, you now have the skills and the teams to be able to do it in house as well.

Right.

Which is something I think you and I have done quite a bit in previous lives.

Yeah.

Yeah, that makes sense.

Yeah.

Yeah, that, that, that is so obvious of an answer when you actually say it that way.

Yes, that's really well said.

I, yeah, thanks for reminding me of that.

I, I'm kind of like shaking my head like, yep, that's just pretty much how you should do it.

Yeah.

All right, well then with that said, I'm going to put your feet to the fire then.

Since you just flipped my head again the second time in this podcast.

Sam a. Sam a CEO and or Sam even a chief supply chain officer.

Whatever you want.

I'm listening to this episode as many of them do what And I'm, I'm align on my strategy.

I want to get going.

What step should I take, Omar in the next 30 days?

First step, be clear about the objectives and the outcomes.

Right.

Objectives and outcomes.

Right.

So, you know, again, I think number one, we go back to what we said previously.

It's not about applying technology to apply technology to stand up.

And in some cases it is like, let's be clear.

Some, a lot of our retailers have boards where, you know, you need to say, hey, I've got these four agentic use cases because it's the, it's the buzzword of the day.

Right.

But at the same time, we also have the responsibility to make sure that it's meaningful towards the organization.

So be very clear about your strategic outcomes.

Right.

Identify the areas you want to go after and be clear about.

Again, you know, we hit about, you know, we hit on it quite a bit.

The ROI of what I'm trying to solve.

Right.

And look at what I would say the lowest hanging fruit first.

Right.

So where's my order?

The things that the address talking about, these are menial tasks that people are doing.

You can get quite a bit of benefit just out of looking at the lowest hanging fruit.

First, I think we all have a problem.

The first problem is applying technology for technology sake.

Second is we have a tendency to over complicate what we need to do.

Right.

And in many cases the answers are right in front of us.

And so I think the biggest thing for a supply chain executive is take a step back and look at the end to end and say, okay, look, where are my big problem points relative to.

Again, I go back to what I was saying.

Labor.

Right.

Status.

The connection points between that are disconnected today.

What if I could connect?

Right.

So, you know, I think they need to really do an introspection to say, look, you know, how do, how would I really like this to work and remove some of the constraints that we've had over ourselves, right.

At least from a strategy perspective and then get into okay, right.

What, what are the key kind of metrics that are always, that I can achieve if I connect these two things together?

How, you know, you got where I'm going, right?

To me, I think the, the key is really looking at the supply chain end to end, looking at your, your lowest hanging fruit because in many cases it's right in front of you and then being able to kind of align again, you know, three to four objectives, right.

That, that you're going to, to, to, to hit in the near term and then focus the entire organization around them.

Right.

All right, well, let's close with this.

This has been a really riveting discussion.

I wish I could spend even more time with both of you.

Final question for both of you.

Now I want specifics here.

This is my kind of confession question.

If I could only do one thing as a retail executive after listening to this episode, if I could only implement one single highest impact action, what would it be?

Is there an area I should focus on first?

Is there a way I should structure my team?

What is it?

Adil, why don't you go first?

I would say automate visibility.

You know, if you can't meet promises to your tier shoppers, that's your first line of defense.

And that's actually one of the highest ROI use cases for agentic AI Automate visibility.

Explain more about what that means for those maybe that are unfamiliar thinking about

that order Lifecycle Journey.

So the moment the shopper checks out to the point that they're, they receive the final delivery, mapping out what are all the steps involved in here?

Who's touching the product?

Both digitally, physically, what are the dependencies?

And then deploying AI agents at each of those dependent steps and then exactly as I mentioned before, connecting them together into a unified way for you to act.

So when your shopper comes in through any channel, voice, email, text, your, your, your website, chat and says, where is that order?

You have an answer.

Oh my God, so good.

Omar?

Me too, right?

For sure.

And I think again, the second part of that is think of all of the configuration and management that you need to do around that, right.

To keep the systems right.

So that you're promising the right promises and you're so again, think about all of the, when you're trying to connect the ecosystem, right.

Whether it's supply chain execution or whether it's around the customer, etc.

All of the data that is required to configure.

You remember Chris, in our old days of you know, hey, I need to change this eligibility and that eligibility and I need to change this.

Oh no, this thing is going viral now.

I need to be able to do this, that look across your workflows and configurations.

And I think that to me is, you know, I think add this spot on.

And me too, that's probably where I would start.

Where, where's my stuff deploying agents and not just where's my stuff being able to actually action them and, and, and do things like replacing an order, changing an order, you know, doing cool things through the lifecycle to make the customer and the company whole.

The second part that I would come in with is also how do you actually set the right data so that those promises are being set the right way so the configurations, the workflows, the connective tissues that you want to be connected.

I think that's something that companies spend a lot of time, money and resources on.

And that's an area again, you know, whether you think of it as cost to serve.

Right.

Like how, how much labor does it take me to, to serve my channel.

If I can actually deploy agents to help reduce that cost to serve by automating workflows, automating configurations, ensuring I'm putting the right promises in play and I can deploy agents to do all those.

That would be a game changer.

Yeah.

As I'm sitting back here, I'm kind of like, yeah, that's definitely the answer.

It's like you're available to promise the visibility because you humble things are triggered off of that.

But I'm curious too.

Is there one follow up, quick follow up question.

Is there a metric that people should be tracking in their day to day progress against that area of specification?

So let me, let me go and I'm going to hand it to you.

I think first think about the number of hours that people spend today on answering those phone calls and then with these agents, you know, the reduction that that will resulted.

That's the first.

Number two, think about all of the ratings that customers will provide about delivery and service, etc.

And how that could increase by, by making it better.

Number three, think about, you know, the number of hours and the staff and, and, and, and the people that you need to do all of these configurations.

Right.

That you now potentially you know, can, can, can reposition to, to better, you know, and more, More, more skill tasks.

So I, I think the roi, when you start putting this all together, starts to present itself, and the metrics, the dashboards become also something that, frankly, you know, we can, we can make into a reality with this new technology to actually measure your customer experience, measure your internal experience, and measure, you know, the cycle time that it takes to do things right.

Couldn't agree more.

All right, well, so I imagine there's going to be a lot of folks listening that are going to want to pick both your brains, as I do actually want to have you back on to continue this conversation at some stage, because it was really enlightening for me.

There were a couple aha moments where if you're watching this on video, you could probably see them on my face.

So if people want to get in touch with either of you, what's the best way for them to do that?

Adil, find us at NPS on our website, www.infios.com or reach your reach your favorite sales rep. Well, like I said

before, thank you both for this wonderful conversation.

It's been incredibly valuable, and I appreciate your time.

And Omar, it's always good getting to rap with you in front of our audience as well, because you've really educated our audience on.

On what I think it takes to at least think about, if not succeed in this new era of agentic AI in commerce.

Of course, today's podcast has been produced by the great Ella Sirjord.

I am Chris Walton.

This has been Confessions of a supply chain executive.

And never forget Omnitalk fans, Confessions are almost always good for the soul.

Be careful out there.