Orchestrate all the Things - The Skill, Not the Stack: A Decade Ahead on Decision Intelligence. Featuring Fred Laluyaux, Aera Technology Founder & CEO

Episode Date: July 23, 2026

Talk about what the technology can do, not how it does it. That's the decision Fred Laluyaux made with Shariq Mansoor when they founded Aera ten years ago, and it's still how the company talks about its decision intelligence platform today. Decisions sit at the center of what Aera builds. Aera automates, augments, and digitizes the thousands of decisions companies make every day to run their business. Aera pioneered the decision intelligence category years before generative AI gave the industry a vocabulary for it: agents, skills, memory, decision traces, context – graphs even. From the start, Aera was conceived as an agent, in the spirit of the digital assistants of that era. Laluyaux championed the word "skills" years before the major AI players made it common currency. He co-founded Aera in 2016, before GenAI and large language models, and has served as CEO since. In this conversation, Laluyaux traces the arc from DJing in Paris to building SAP's finance line of business to Aera's decade-long bet that decisions -- not dashboards -- are the unit that matters. And why, he argues, letting AI take on more of the decision doesn't have to mean losing control. Laluyaux's analogy: he trusts a Waymo with his kids more than an Uber driver because he can see exactly how each choice is made. That is what decision intelligence is really about.

Welcome to orchestrate all the things. I'm George Anadiotis and we'll be connecting the dots together. Stories about technology, data, AI and media and how they flow into each other, saving our attacks. Talk about what the technology can do, not how it does it. That's a decision Fred Laliob made with Sharik Manchu when they founded ERA 10 years ago. And it's still how the company talks about its Decision Intelligence platform today. Decisions sit at the center of what ERA builds. ERA automates, augments and digitizes the thousands of decisions companies make every day to run their business.

Errol Pallneed the decision intelligence category years before Generative AI gave the industry a vocabulary for it. Agents, skills, memory, decision traces, context and graphs. From the start, era was conceived as an agent in the spirit of the digital assistance of that era. Lalia championed the world's guilds years before the major AI players made it common currency. He co-founded era in 2016, before Gen AI enlarged language models and has served as CEO since. In this conversation, Laleo traces the arc from DJing in Paris to building SAP's finance line of business to ERA's decade-long bet that decisions, not dashboards, are the unit that matters. and why, he argues, letting AI take on more of the decision doesn't have to mean losing control.

Lallius' analogy, he trusts a way more with his kids more than an Uber driver because he can see exactly how its choice is made. That is what decision intelligence is really about. Part of the global vision for ERA was, and it starts with a very simple question that I ask operators of businesses today all the time. So you have all this great system, George. You've deployed, you know, analytics and data lakes and ERP and planning and all these great tools. Where do you store the memory of the decisions that you make every day?

And then people look at me like, well, well, we don't. And I said, so how can you continuously improve the quality, the velocity, the impact of the decisions that you make every day if you don't know what they are? So you built your business with policies, enablement, process, tools, norms, hoping that people will execute the playbook exactly at the right time and it doesn't work. So the vision is you've got to bring that decision memory, the memory of all the decisions that you make at the individual level, at the departmental level. How do you operate as a revenue generation department versus the supply chain versus and at the enterprise level?

That memory is the foundation of everything. And a skill will feed from that memory using machine learning within different technology. So that was our vision from day one. I hope you will enjoy this. If you like my work and orchestrate all the things, you can subscribe to my podcast available on all major platforms. My self-published newsletter also syndicated on Substack, Hackern, Medium, and DZone, or follow, gesturate all the things on your social media of choice. All right.

So, thanks for having me, George. My name is Fred Laloyo. I'm the co-founder and CEO of ERA. So the story of my life in three minutes, first of all, it's interesting you ask that question because that's the question I always ask when I interview people. Born and raised in Paris, started my kind of a bit of a nerd. kid, lots of interest in sports as well, went to college in Paris with a very clear mind that I would make my way to the United States. I don't know why the bug got me, but the bug got me

when I was a little kid. Spent four years in Paris, my first passion and my first business was actually being a DJ and playing music and doing light shows and setting up stages. I love that when I was in college, it got me through. But the bug of tech got me quite early, and I did my first internship in the US when I was 20. Got a chance to come to Milwaukee, Wisconsin and work with a great little company doing technology stuff, and the bug got me there.

So one thing led to another. I started my first tech company when I was 23 in Paris, before the word startup even was a thing. I spent four years building a really cool technology that ended up being a disaster because a thing called the internet, that ages me now, came along and wiped the technology that we had built. So first major disruption that taught me a lesson and decided to move into software.

So when I was 27, I started building the French and the European operations of a software company from the UK doing enterprise software, cost management. It really focused on the technology called activity-based costing or a methodology, I should say, called activity-based costing. build that quite successfully, got the opportunity to move to the U.S. when I was 32, moved to Atlanta with my wife and took the American operations and the global operations of that company. I started building a family in the United States, spent 10 years in Atlanta along the way we sold that business to a company called business subjects, doing analytics, spent a couple of years with them, helped them build their performance management. business, then we got acquired by SAP, spent a couple of years doing a super interesting project, which was the integration of SAP and business subjects, which was really, really interesting.

Then I built SAP's finance line of business. At the time, you're looking at, you know, 2008, 2009, 2010. SAS became a thing. The industry looked at it by, by specialized by line of businesses. So I was the opportunity to build a finance line of business. And And then I left in 2012, built a company called Anaplan that became quite successful with a great team. 2016 left to build ERA. And it's been basically since 2017 deeply, deeply passionate about ERA's story and what building here. Live in San Francisco, three kids. Some of them are grown up now.

And one of them lives in Africa and work for the Peace Corps. The other one has moved to L.A. and still have a son at home, 17-year-old, who is going to be a senior in high school. That's kind of a quick story. Okay, cool. So I already identified two things we have in common. First, a passion for music and DJing.

I also do that as a side gig. Ah, good, good. And second, and actually I went through a kind of similar, let's say, soul-searching moment around the same age that you did. Like, okay, do I want to do it as a job, or should I keep it as a side thing? I think the answer to the question is obvious now. And second common thing in common,

I identified this 10-year anniversary. So you started, you went away to start your own thing with era at 2016, and I sort of did the same with my own things, but actually- Excellent. So actually the focus is you, not me, so I won't talk about what I'm doing. The point is...

I want to hear more now. No, actually, I can tell you more, you know, after we finish the recording. But for now, what I want to focus on would be actually what it is that you do with ERA. And more specifically, the category that ERA is. And I'm just saying that you could as well say that ERA has actually helped define and save that category, which is called decision intelligence. And I know that many people will be familiar with it because, the gardeners of the world have been actually including it and analyzing it for a while now, but not necessarily everyone is familiar with that.

So I'm going to ask you to very quickly walk us through decision intelligence, and how do you approach that? So quite simply, the definition of decision intelligence is the digitization, how do I automate, augment and digitize, the decisions that companies make thousands of times a day, a week a month to actually run their business. You have in a big company or a small company, you have operators day in and they out. When they're coming to the office, they have to make sure that the right product is being made in the right quantity, in the right format, at the right place, optimally shipped.

It has to be getting to your customers at the right time. It has to be priced properly. You have to create the right promotions. You have to buy the right quantity. you have to, all these decisions that you make to run your business, well, they can be digitized today. And this has been our vision for the last 10 years. This is why we built era. We thought that there was an opportunity to move from people making and executing decisions, thousands of time, repetitive, what we call the dull repetitive work, and shift that to having AI augmented system actually be able to do a lot of the work. And there's a lot of benefits when technology does it but the category that we have created you know it's decision intelligence when we started and we

talked about the self-driving enterprise we talked about the autonomous enterprise which is a big theme today a lot of large organizations are talking about autonomy and and it's fundamentally what decision intelligence is so we we pioneered the category we kind of probably came up with the with a lot of the the concepts that are being very popular today and i think today we're leading the category but high level digitization of decisions is what decision intelligence is all about okay okay thank you so actually the reason i know a thing or two let's say about decision intelligence and also about what you specifically do is because this is not the first time that i came across era so i've had the pleasure of connecting with your ctio sorry sarik mansure back in

2022 and actually for the purposes of revisiting this conversation to refresh my memory and to get a little bit of background for our conversation i couldn't you but notice that the description of the way aira approaches decision intelligence and its platform seems a lot like how many organizations today describe their vision of using lLM-powered agenda to organize their operations and I mean to make it more concrete. There's the layered architecture, so data, intelligence, and engagement. And also the decision cloud, which I bought, you named as an engine to digitize human decision logic,

and even the use of skills. That one got me, I have to say, quite deeply, because I was like, okay, so skills are kind of ever-present. Today everybody's talking about it. It's funny you say that. Equipping your agents with skills, and I was very surprised. I'll be honest, I didn't remember that this particular term was used,

but rereading that, I was like, wait, you were talking about skills like before the whole, before even SATs-DPT actually came into the field. So there's a lot that you were seeing ahead of your time. And my question here is that, okay, so what have you learned trying to solve the enterprise decision-making problem before GenAI came along? Yeah, it's funny you say that because even internally, and I've got a fantastic team, I had to fight tooth and nail. We launched a company with a website that's pretty much similar to what you have today. The message has not changed.

We talked about error as an agent. So you caught the word skills, but era has been designed as an agent. That's why it's called era, like Siri or Alexa, A-E-R-A, four letters, a tool that you can call, that you can convert. with. We conceived that era. You have demos of us doing like regimentary experimentation on a phone and asking questions, how can I improve my inventory levels? And my vision and our vision and Shariq's vision was always that this platform will learn how to perform certain decisions, execute certain tasks, and those are skills. So I've been fighting for the word skills for 10 years, including against our own, sometimes our own team, saying, no, it is a skill. It is a skill. It is

of competency. And what defines a skill is that it evolves over time. So when everybody was talking about apps, we talked about skills because the vision was, and I'm going to get to your specific question on LLM. The vision was always that a skill was going to improve over time. Part of the global vision for error was, and it starts with a very simple question that I ask operators of businesses today all the time. So you have all this great system, George. You've deployed, you know, analytics and data lakes and ERP and planning and all these great tools. Where do you store the memory of the decisions that you make every day? And then people look at me like, well, well, we don't. And I said, so how can you continuously improve the quality, the velocity, the impact of the decisions

that you make every day if you don't know what they are. So you built your business with policies, enablement, process, tools, norms, hoping that people will execute the playbook exactly at the right time and it doesn't work. So the vision is you've got to bring that decision memory, the memory of all the decisions that you make at the individual level, at the departmental level.

How do you operate as a revenue generation department versus the supply chain versus and at the enterprise level? That memory is the foundation of everything. And a skill will feed from that memory using machine learning within different technology. So that was our vision from day one. Now, we built this entire platform without LLMs at first because LLMs were not available. That dynamic intelligence, that reasoning capability was not available. So how did you do it? Well, we normalized everything.

We used deterministic logic. We learned by using machine learning, and that worked. But it was slower, harder, and it still is. It delivers incredible value. It's incredibly robust. It's incredibly trustable. But all these techniques are slower. they're not as flexible.

So you bring LLMs, the power of the dynamic reasoning power of LLMs, and you open up the ream of possibilities of the platform that you built. You open it up to unstructured data, which we were not really, really good at before. You open it up to dynamic reasoning. I talked about learning from the memory. Today you combine the machine learning that we have developed over the years. That's still hyper-relevant for high-volume. high density recommendations.

But you complement it now with dynamic reasoning and I can learn from just a lot less data. So this whole addition of LLM in the platform that we've built end to end with data, intelligence, automation, and engagement, turbocharges, bring a layer of flexibility and power that we didn't have before. But the fundamentals are still the same.

Okay, so then would I be right in saying that it sounds to me like there is a deterministic, so to say, element at the core of your engine. Yeah, 100%. A lot of the decisions that you make every day are completely fine with deterministic logic. If this happens, then if I identify a risk of stockout, I have five options. I can manufacture more. I can expedite inventory. I can transfer. I can let my customer know that I have an issue. I mean, those are deterministic rules. They don't need agentic logic every time. When I'm managing a much more complex problem than I'm going to need a combination of potentially deterministic and agentic. And this is the power of the platform. And also, you think about the power of agentic, it's incredible, but it's

slow and it's expensive. You want to use it when you have to use it. Think about agentic at this incredibly smart employee that you have. that is incredibly expensive. You don't want that person to be working on mundane repetitive work. You want that person, that PhD in math, to be working on the most complex optimization problems that you have. And you won't pull that person for every single like, hey, you know, if this happens, then that, then do that. No, that's not what you want to do. Think about the technology the same way. The most important thing is to have a palette of capabilities. And when I talk about deterministic, where you're probably picturing a workflow, right, a set of conditions,

but there is more to this, right? When you make a decision, you deploy a logic, we call that agentic, ambient intelligence. It's ambient. It's running all the time. It's agentic, but it's also deterministic, and it's intelligence and its ability to determine the best option to a specific problem, and as I said before, learn from it. But you also need to project, to predict, to optimize, to allocate to select. And a lot of that, a lot of that calculation is not agentic at all. It's a good old multidimensional cube that you need to do your projections in your plan. It's a machine learning model. It's a statistical forecasting model. So the key to making this whole skill capability work is a combination of the data model that contains your data. So you can do it directly in your ERP.

You can bring it to error. Doesn't matter. It's that memory. that knowledge of how to make decisions is the intelligence, deterministic, probabilistic, agentic, hybrid. Doesn't matter. You have to be able to select the best way to run your decision. It's the calculations that are required to support the decisioning intelligence. And then it's the execution. You need the capability, the platform, the decision intelligence,

to execute your decisions. Otherwise, George, what do you have? A very noisy employee. You have someone who tells you all day long. I recommend that you do that. And if you don't have the ability to capture the decision and execute it, then your memory is useless because you have a memory of recommendations, but you don't know if the action has been taken, so you can't use it.

Right? So that loop from data, events, sensing to decision logic, which can be simple, deterministic, complex. multidimensional, blah, blah, blah, blah, the calculations that you do every day to actually support the decision, the engagement with the users, do I need you to tell me what you think or can I do it on my own as the software, and the execution and the memory and the learn, right? We talk about understand, recommend, act, and learn. And a lot of tools today are super good with one aspect of it. But if you bring, if you want to do decision intelligence at scale, you have to bring all this capability into a single connected ambient environment as a brain right we've been talking about this forever so imagine that you're really good at calculating i'm really good at reasoning someone is really good at acting and someone else is really good at learning and you try to get the four of us to actually work very very fast it's going to break you have to bring all this capability into a single environment and then you have to enable that that environment to scale we work with the largest companies in the world

world. So you have to bring this intelligence, dynamic intelligence, on top of your transactional data, your planning data and system. When I say data, I mean data and tools and system, and make it scale. And in the world of enterprise software, we were going back 10 years ago, complexity, volume, and speed don't work well together. You usually, if you think about all the software, you have my software that is really, really good at managing complexity. And if you load a lot of data, it's going to be very slow. And if you have a software, if you take a simple problem and a large volume of data can work, but combining those three things, volume, complexity, and speed, it's very hard.

So when we launched era, and that was the inspiration, we looked at consumer software. We looked at software that was born in the web. I asked myself the question, how does Google do it? How does LinkedIn do it? How does Facebook do it? because they manage massive complexity and it works at incredible scale and speed. So we kind of use some of those architectural design principles. We shariq, you talked about Shari Kohl there, is brilliant engineer, to actually crack the nut.

And 10 years in the journey, I think we've cried the nut. And now we're going from pioneer to leader. I think we're the leader in decision intelligence. And now, really, the next milestone is to become the standard and we're working toward. But I'm digressing a bit here. Well, it was an interesting digression nonetheless. So I think it's very much excused, if not welcome. So he touched on a few points I would like to follow up on,

but let's see where to anchor it. I would actually anchor it to the last of those points. So the trip down memory lane, let's say, to the enterprise systems and enterprise architecture. And to tie that also with the fact, well, we talked a lot about adjacent. genetic AI, even if not by name. Because everything you mentioned about the determinism and how,

and to me, what separates AI or Gen. AI or LLM from agentic AI is the fact that you also highlighted. So not just recommending, not just generating text or predictions or whatever, but actually executing. This is the crucial line. And I think the people who cross it actually feel it. So here's the thing with the process it. proliferation of agents, so the conversation is actually shifting now to decision traces.

So where do these agents live? How do they integrate, how do they align with each other, how to use them, and who owns them? And at the same time, using LLMs to support retrieval via some patterns known as rugs or retrieval augmented generation and graph rags or graph-based rug. It has made something that I care about and I have, let's say an affiliation for knowledge graphs it had made them relevant for audience that go beyond the historical applications of enterprise models graph so what i'm trying to get at with with all of that is the concept of context graphs and it's something that's relatively new it was just introduced at the end of 2025 beginning of 2026 but it's something that at least in my world i see

kind of blowing up and everybody and you can tell because everybody is saying what we have context graphs too and of course everybody means something slightly different so i want to anchor it to to forester actually to one of foresters uh key people charles birds who's a vpheus forester and charles argued that context graphs are a convergence and not an invention and by correct what he meant by that is that you can identify decision traces in a wide range of disciplines and tools, and most notably enterprise architecture. And that caught my attention as a former enterprise architect myself. And so what I argued, what I tried to bring to this conversation,

is that we actually need to go beyond the decision trace and to leverage knowledge architecture for context graphs. And I know that you also leverage complex graph analytics, and you combine infrastructure for graph vendors with proprietary implementation. So I wonder, what's your take on all that on the whole graph slash context graph take? The way we crack the problem, I don't want to tell you too much either on this one, but I'm going to go back, I've got to be careful because I can get carried away here. But I always go back to the basics, very, very, very, very, very,

very simple. At the core of error, forget about the technology layers. At the core of error is the ability to memorize the decisions that are made. Let me get back to, forget about the technology, just look about the logic here. What is a decision? There's the first question that you ask yourself. What is a decision? How do I capture a decision? It's in your head. The way we approached it. And I mentioned earlier our inspiration from Google back 10 years ago. Forget about context graph, forget about agenting, just go back to the core here. We thought, how did they do it? How did they train a search engine?

They'd be testing. They gave you a list. You ask a question and they give you a list of answers. And remember back in the days you had the time it took to provide that answer, you know, 2.3 seconds. Based on your interaction with that page, the system would learn whether the answer was the right answer. to your question. And that when you actually don't come back to the page, it's a decision.

You're actually deciding, you're informing the system that the answer was correct. And of course, there are a lot of shades of gray around it. So we took the same approach and we say, well, what can we do? There is a set of enterprise data. We can build that ontology. We can sense the signals

and either on schedule or based on specific events. We can deploy your decision logic based on your data to deliver a recommendation. That recommendation, and now I'm going to the context, that recommendation as the context of what triggered it, whether it's a difference in inventory level or a signal from a client, something that came through an email. That context is memorized in an object called a decision object. So the recommendation understand the context of its creation, and it memorizes it. It also understands the different options that were available.

It memorizes the different options that were available, as well as the expected outcome. And keep in mind that this is not a single dimension. This is multidimensional. The impact of a decision will be felt across multiple dimensions. In supply chain, you will think about cash cost, service level, carbon, water, whatever it is. All that information, that contextualized information,

is stored in an object that created for every decision, the decision object, that then populates the decision memory. Based on that, the system will calculate a confidence score. And that's an algorithm that will calculate the confidence level that if you accept the recommendation, the expected outcome on the primary decision will be delivered. That allows you to define some business rules, that allows you to say,

for all those decisions in the context of these customers or these products or whatever it is, if my confidence level is north of 90% execute automatically. That's how you get to the level of automation that our customers are getting to. But maybe for exceptions or confidence level is lower, go to George proactively and say, George, I have this recommendation for you. These are all the analytics that supported. These are the conditions. This is the logic.

These are the options. This is the impact. Maybe I give you the ability to simulate, to test, and then you decide. That decision gets captured. Your interactions with the tool get captured. So that's one source of building the context graph. The other one is new, is the ability for you to go in error and say, error, how can I improve my service level without impacting my working capital for this customer in that area?

And now you engage into a dynamic reasoning with your tool, fully leveraging the power. of LLMs and the memory that you build for your industry, your enterprise, your department. That can lead to a decision. So now the system has to memorize your thread with the tool and enrich the graph with that unstructured, that situational decision. And that graph at the end is what is now populating and leveraging both your structured and your unstructured decisions. And that's what populates the graph that then allows the system to learn.

And then where we are right now, which is super interesting, is based on that and based on the awareness of what the outcome is relative to the expected outcome, the system can now come back leveraging at LMs. That's another area where you leverage agent to make policy recommendations. So now it comes back and say, George, you know, I've analyzed all the decisions you guys are making on this. And it appears that every time you have to ship by air, you have to go in front of this person and you have to do this and that delays you. And I know that maybe you should change a policy. So now you have a system of intelligence that leverages that contextualized graph to

generate policy changes. One of our largest customer, the system last month generated 20, sorry, 30 policy recommendations. They've accepted 28 of them. And that's an, and And all the way to do it would be to bring everybody, the thousands of people who are making decisions in the company, and say, okay, let's go into a big room and let's try to analyze objectively how we should improve our policies. Imagine that's not working, that's broken. But when you have that memory of decisions, you lend it.

That digital intelligence work for you. And you can now have the, policy management at your fingertips. Every time you change your policy, you know it's being executed because it's applied to all the decisions that you make. So that's it. This is the way to, and it's 10 years.

It took us 10 years to get there. But this is fundamentally the way to remove the human-based inefficiencies from how the companies make and execute decisions. This is the unlock that we're looking for to build a more sustainable and more efficient to enterprise world. This is what we needed. And I think we're getting there. And I again, digress quite a bit from your initial question. But I wanted to put the question that you asked,

which is technical in the context of the business outcome, which is ultimately what you want. Cool. So if I may ask a follow-up question or actually, you can consider a question or a comment. It's up to you. So based on what you said, I would say like the takeaways, probably I would summarize it as well. It sounds like you've actually had a context graph for longer than most of the people that actually say they have a context graph today, but you just don't talk too much about it. Sometimes I feel like we get, we fall in love with the technology and we forget the business outcome. I'm really focused on trying to drive business outcome. I work with boards, executives, CEOs, CFOs, super supply chain officers. They don't care about this technology versus that technology.

I think we really focus on the business outcome. Now, you engage with Sharik and you're engaged with our tech folks. I mean, of course, you know, when we have our engine, today I'm going back to Mountain View and we have a four-hour product review and we're going to talk about this. But from a messaging standpoint, we're trying to push the business outcome. What Hershey's been achieving, what, you know, Dell is. achieve where all our clients are achieving because ultimately this technology serve a very simple purpose, remove the human-based inefficiencies from how you make you ship. We have an incredible

use case with a Western Governor University, the largest digital university in the United States. They deployed era, not for supply chain, for student retention, to help engage with the students at the right moment before they would drop. And they've talked about it publicly. This is why I can say that in this session. The results are amazing, right? Because now you bring the right information in the right context to the right person at the right time and you get an outcome, which is a student that says, I'm not going to drop off, I'm going to drop out, I'm going to stay connected to the course that have elected to follow. And those results are real.

So technology for the sake of technology is not what we're trying to do here. We're leveraging technology. We were super early. We talked about agentic. We talked about skills before LLMs. But we don't really care. We did that in the context of a very clear vision, which is how do I improve the quality of the decisions that we make every day?

And that's pretty much what's driving us. Maybe we should talk more about the tech and the under the hood. You know, 10 years ago, I made a decision with Sharik, that we would not talk about how we make the decisions. We would talk about the skills themselves. To go to our website, there was very little on the platform and a lot about the skills. Here is how you optimize your inventory. Here is how you optimize your transportation, your procurement.

We talked about the business outcome, not so much about the technology underneath it. And the vision was that as we released new capability and we do every month, we talk to our community of engineers and the people or the customers. They know what's happening in the background, but the public is more concerned about what it does. for them. You think about Siri or Alexa. You care about the fact that Alexa can now turn the lights in your house and play some music back to our DJ discussion. You don't really care so much about the fact that they've added a new context graph in the background that allows you to memorize which playlist you like. You want to talk about the skill that Alexa and I can play music as well as turn on the lights. And those are competency skills that it didn't have before.

And I care more about positioning the skills that will benefit the enterprises that we work with in the market with all partners than what's happening underneath. Fair enough. Okay, so I think we're getting close to wrapping up. So so far we've talked about the past and the present and with a little bit of hint towards the future. So let's wrap up with focusing on the future. So I read about the introduction of agentic reasoning for enterprise decision in ERA's platform. And I think to tie that back to what you just said about what your focus is, the technology versus the business layer, let's say. So my question is this basically.

So what kind of, be you see any kind of risk or pushback by introducing? basically autonomous decision making. If I'm a business, if I'm a business owner or business stakeholder, my confession to you is that I would feel a little bit skeptical, if not threatened by that. Because to give away agency, I think this is the key word here to an AI agent,

means that, well, I have less control over the decisions that are going to be made. So how do you really? respond to that skepticism? I hear it, and I will say the following. And I do that all the time. So you're an executive, okay? You are in charge of big companies, small company, doesn't matter.

And you have hundreds, dozens, thousands of people in your company that are making decisions every day. Do you actually understand what they do? You think you do. You expect them to be able to. way, but you have no idea. How they actually make decisions. Are they making them on time? What are their biases? Are they optimizing for something that maybe is not what you want them to optimize for? And in a very fluid world where things are changing all the time, where disruption is

the norm, you have to, you're not in control at all. You define policies, you have software, you have controls, you have blah, blah, blah. When you deploy decision intelligence, you actually now can look at the screen, it's called a control room, and see in real time what decisions are being made with the combination of technology and people. So you're not losing control. You're gaining access to the real-time visibility on how your enterprise decides. Today, you don't see it. You see the consequences of the decisions. You create a plan. You roll it out. You hire great people, they usually have the same profile because it's easier to drive the army in a certain way if you have the same type of soldiers. And then you control at the end, did we achieve the goal?

But you don't know what's happening how the sausage is made. D.I. Decision intelligence is the opposite. You see in real time day by day, second by second, is that decision delivering the value that we're optimizing? And then your business operators, they are actors of the process. They see, they act, they provide their input in the decisions that are being made. So you gained visibility. You don't lose visibility. And we talk about decision autonomy. Of course, your enterprise decides autonomously as much as you decided for it to do so. You define the parameters, you define the controls, you define the guardrails. And it all goes down to the terministic versus a gigantic logic that we discussed earlier in our conversation.

If you're relying 100% on agency, you're in trouble because it's going to be incredibly expensive, slow, and setting up all the guard wells might not be relevant or working all the time. But by being able to combine the right technologies to optimize the way decisions are being made, whether it's deterministic or whether it's egentic, do you have the guarantees that...

So we use Agenic to actually configure the deterministic capabilities in the platform. We use Agenic to quickly configure the deterministic capabilities in the platform. So you know that it's going to run as a computer, not just as a delusional or a, you know, capability. So you're in full control. You gain control. You gain access to that new data. You build that memory of decisions that you've never had before. that allows you now to optimize across your value chain,

which you cannot do today when you're relying on people. The number one pitfall that we've seen over the last 10 years when we help our clients deploy decision intelligence is the lack of true understanding of how decisions are actually made. You'll ask those executives in the company, how do you do this? Well, this is how we do it. Then you deploy the software. and then you have operators in a warehouse,

in a production center, in a mixing center. They're saying, sorry, guys, this is not how we decide. Oh, we thought we knew. So that's the number one people that we've learned. Now it's getting better. The second one that we've learned as we wrap up here is we've seen an evolution, a rapid evolution from, I need my operators to validate every decision that the system makes

to I'm going to get my operators out of the loop. because I know that every time someone touches a forecast number, it degrades the performance of the forecast. And the analogy I use is, I live in San Francisco. With all my kids going around, I'd rather have them be in a WEMO than being an Uber. And not disrespect to their Uber drivers, but I don't know who they are.

I don't know how they think. I don't know what they've done. I don't know if they're tired. I know they work for 15 hours before they get my son to his party. I know the WEMO is going to work. Following the rules, I've got full control. I can track this. And this is what, and again, five years ago, 10 years ago when this is started, and there was no way I would let my son, oh, my daughter is

going away more. I wasn't, I didn't trust the system. Now I trust it more. And apply that to repetitive, normalized decision logic. And you have the answer. Thanks for sticking around. For more stories like this, check the link in bio and follow link data orchestration.