Sep 11, 2026 · 28 min · 11 segments
\*Hosted by Andrei Palamariu* I’m joined by Uzair Bawany, CEO of Oii.ai, to explore what happens when AI moves beyond the hype and into the operational reality of global supply chains. Drawing on his…
So, When you think about everything happening with agentic AI and slapping AI and everything left and right just to make it more shiny, how is AI working in this mix and how is old-fashioned mathematical engine solvers working hand in hand or how do you distinguish between the two?
Well, I think the key point is, you know, there's this whole concept of black box, right? Can you explain the technology? If you can't understand it as a buyer and your vendor can't explain it to you, then what are you doing trusting it is my question to anybody.
So we have a very, I mean, what we're solving is that challenge of cost, service, and cash.
What is the most optimal way that trial can be solved? And that is a mathematical equation.
And we can actually prove, and we actually had a client, a big retail group that wanted to actually understand the technology.
And we went through every single code of algorithm in order to explain how we derived with two of their PhDs.
So if you increase your safety stock by 10 days, what's the impact of your service? If you reduce it by 10 days, what's the impact? So you can see that balance, that mass coming into it.
The concept of, again, this agentic AI is a word that everyone throws about now.
And that then flows into one person, one individual, one agentic person who actually becomes the question and actually operates that fleet.
What you do is you ask the question to the machine, where's the most optimal value I can get on my safety stock on these particular SKUs on this month? And it will go in, it will fire out questions to the agent and come back with an answer.
That's what agentic is about, managing the agents to get the optimal value for that discussion.
And what we've done is we've built from the bottom up, we've built swarms of agents in each part controlling it.
And in fact, we've now got a client, it's a large business in the US, where they actually want us to monitor certain levels.
So we can build an agent to monitor certain things to then alert the agentic person, the controlling agent, to give that view.
So agentic AI effectively works when you've got the infrastructure of the agents, and actually it's doing the work.
It's not going out to the outside world to ask a question and coming back with a chat-chief-y-ty response.
So, When you think about everything happening with agentic AI and slapping AI and everything left and right just to make it more shiny, how is AI working in this mix and how is old-fashioned mathematical engine solvers working hand in hand or how do you distinguish between the two?
Well, I think the key point is, you know, there's this whole concept of black box, right? Can you explain the technology? If you can't understand it as a buyer and your vendor can't explain it to you, then what are you doing trusting it is my question to anybody.
So we have a very, I mean, what we're solving is that challenge of cost, service, and cash.
What is the most optimal way that trial can be solved? And that is a mathematical equation.
And we can actually prove, and we actually had a client, a big retail group that wanted to actually understand the technology.
And we went through every single code of algorithm in order to explain how we derived with two of their PhDs.
So if you increase your safety stock by 10 days, what's the impact of your service? If you reduce it by 10 days, what's the impact? So you can see that balance, that mass coming into it.
The concept of, again, this agentic AI is a word that everyone throws about now.
And that then flows into one person, one individual, one agentic person who actually becomes the question and actually operates that fleet.
What you do is you ask the question to the machine, where's the most optimal value I can get on my safety stock on these particular SKUs on this month? And it will go in, it will fire out questions to the agent and come back with an answer.
That's what agentic is about, managing the agents to get the optimal value for that discussion.
And what we've done is we've built from the bottom up, we've built swarms of agents in each part controlling it.
And in fact, we've now got a client, it's a large business in the US, where they actually want us to monitor certain levels.
So we can build an agent to monitor certain things to then alert the agentic person, the controlling agent, to give that view.
So agentic AI effectively works when you've got the infrastructure of the agents, and actually it's doing the work.
It's not going out to the outside world to ask a question and coming back with a chat-chief-y-ty response.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.