- 2 months ago
As AI systems autonomously orchestrate global supply chains and smart factories optimize production without human intervention, manufacturing stands at a defining crossroads. The European Union's Industry 5.0 vision prioritizes worker wellbeing and human-centric automation, while efficiency-first competitors race toward full autonomous operation. This panel brings together leaders from logistics, policy and innovation to debate whether human-centered values represent the future of competitive advantage—or a costly compromise.
Category
🤖
TechTranscript
00:00All right. Thank you. Good morning, everyone, and welcome to our panel. The autonomous future. Can industry 5.0 compete
00:12when machines run the show? I'm Kelsey Chang from Caixin Global, and we are a finance and business organization, news
00:19organization based in China.
00:20We are now entering a new phase of industrial transformation. AI is now going beyond just summarizing data or automating
00:31tasks. It's increasingly helping companies run operations in real time and making decisions on behalf of humans.
00:39Now, that raises a very big question for the future of industry. What is then the role of humans? And
00:47more importantly, can Europe's Industry 5.0 vision, which emphasizes human centricity, resilience and sustainability, compete with models that focus
00:57on speed and cutting costs?
00:59Today, we're very lucky to have four fantastic speakers to help us break down these questions.
01:05First, we have Olivia Amodzik-Belit, Director of Strategy and Industrial Transformation at La Post. Olivia brings an operational perspective
01:15in logistics and services.
01:17Next, we have Julie Teigland, Global Vice Chair for Alliances and Ecosystems at EY.
01:23Julie works with big companies and tech partners around the world, helping businesses carry out transformation at scale.
01:29Next, we have Rudy Kuhn, Lead Evangelist at Solonus. Rudy has spent decades working on process improvement, helping companies understand
01:38how their businesses really operate and how to leverage that information.
01:42And finally, we have Kathy Quashie, CEO of Growing Markets at Inetum.
01:47Kathy has led growth and digital transformation across major companies in tech and telecom.
01:52Thank you so much for being here with us, and let's jump right in.
01:56Now, for the first question, let's set some definitions.
01:59When we say machines run the show, what does that actually mean in your area of expertise today?
02:05And within these AI systems, what is the one function that should remain fundamentally human?
02:11Olivia, can you kick us off?
02:14I would like to start with a very concrete situation.
02:18A few weeks ago, one of our postmen in Lyon, he had to deliver a parcel in an apartment building.
02:26So, he came to the building, which was very hard for him.
02:31He had to struggle to find the address.
02:33Then he rang.
02:35The lady wasn't there.
02:36So, a neighbor stepped in, says the lady was hard hearing.
02:41He rang again.
02:43And then he tried to reach the person and bring the parcel.
02:49So, delivery was completed.
02:51Problem solved.
02:52But imaging one minute the same situation with a full autonomous system.
02:59So far, it doesn't work.
03:02Maybe one day it will.
03:04But so far, it doesn't.
03:05So, postal operators didn't choose between humans and automation.
03:14They've learned, sometimes the hard way, that the question is far more complex.
03:20In France, La Post operates under a universal service obligation.
03:25What does that mean?
03:27It means that we have to deliver every day, six days a week, over 40 million points of delivery,
03:3436,000 municipalities, anywhere in the country.
03:41It means remote islands, small villages.
03:45And that's a tension.
03:48That's the core of our business model.
03:50And that tension is not going away.
03:53No algorithm can solve it.
03:56So, what did we do?
03:59As mail volumes declined, we moved to a hybrid model.
04:03Combining automation and human expertise.
04:07The challenge was to ensure, at the same time, mail sustainability while scaling our parcel's activities.
04:16Today, our sorting centers are significantly automated.
04:21They're processing millions of items daily.
04:24We also run semi-automated distribution centers where you have humans and robots working together, conveyors.
04:34But delivery is a very different story.
04:38In the last hundred meters, reality is messy.
04:41You can have buildings with no access.
04:44You can have, sometimes, addresses that are very hard to find or to read.
04:49You have an expected situation.
04:52That's where customer satisfaction is won or lost.
04:56And no autonomous system has solved it in scale.
05:00This is what we call the last mile paradox.
05:03So, the question for us, postal operators, is not human or a machine.
05:09It never was.
05:10The real question is now, how do we design systems that combine both intelligently?
05:17Because, in the end, automation delivery delivers efficiency.
05:23Humans deliver adaptability, trust, and resilience.
05:28And in a world that is becoming more and more complex, not less, those qualities are not optional.
05:36They're just strategic.
05:38Thank you, Olivia.
05:39We'll dive into some of those solutions a little bit later.
05:42Julie, what do you think when we say machines run the show?
05:46So, I'm going to take a very controversial perspective.
05:51I don't think machines run the show.
05:53I actually think we should be not speaking about such language.
05:58I think, yes, artificial intelligence and machines can support data.
06:04They can run decisions.
06:07They can even execute tasks.
06:09But I want everyone in this room to remember that we need one human quality to stand out above all,
06:16and that's judgment.
06:18Judgment is never going to be replaced by humans or AI.
06:23Whether that's, in your example, the last mile, using judgment for how to do that delivery to that unobtainable address,
06:32judgment on a final position in terms of a very difficult controversy, judgment in terms of what fits best, how
06:42it looks, the mood you want to shape.
06:44You can get support, you can receive suggestions, but humans need to maintain the skill of exercising judgment and therefore
06:56not be in the loop, but above the loop when it comes to leveraging machines and AI.
07:02So, my ask is, I see a lot of great things that can be done with AI and machines.
07:08We should be embracing that, but we should remember, they're never going to run the show, and anyone that says
07:15that is making an excuse.
07:17Thank you, Julie.
07:19Rudy, what do you think?
07:20Yeah, I think that's a really good question.
07:22And, you know, I worked in the automation industry for a couple of years, and I have never seen a
07:28fully automated process.
07:29It's a myth.
07:32All we can automate are tasks.
07:34So, it's about orchestration.
07:36It's about really, you know, looking at efficiency and compliancy end-to-end processes.
07:41And just like many tasks that are repetitive and rule-based, also many decisions are repetitive and rule-based.
07:48And we can easily, you know, have a well-trained AI provided with all the context necessary to make these
07:54decisions.
07:55And we don't want people to work like robots in processes, and we don't want people making, you know, the
08:00same decision over and over again.
08:02And one thing we see of our customers, that if we apply AI and train AI in the right way,
08:07AI can make a lot of these decisions, you know, the typical standard, the routine decisions.
08:12But every time we run into any exception, then people need to talk to people and resolve it and provide
08:20guidance and recreate the process, everything.
08:23So, you know, a couple of years I even wrote a book about autonomous enterprise or contributed to a book
08:28about autonomous enterprise.
08:30And, again, today I would never again talk about it.
08:34It's not about autonomous enterprises.
08:35We should never do that.
08:37What we should do is autonomous operations, autonomous processes, and basically free people from stupid robot-like work, but let
08:45people talk to people.
08:48People talking to people remains key.
08:51Kathy?
08:52I'm smiling because I have a very different point of view.
08:55So, look, I think we're at a generational moment.
08:58And AI is showing us the real impact of transformation today.
09:03And when I think about AI, for me, machines are kind of already running the show.
09:09And hear me out on this.
09:12I'll give you a few examples.
09:13When you think about what autonomous means and what algorithms are doing today, algorithms are managing power grids across Europe.
09:22They are triaging client data, even before a clinician walks into the room.
09:28So, while the machine isn't there running the show in its sense, it's there making a huge impact to our
09:37generational moment in time.
09:39And what's even more inspiring is that when we think about autonomous, autonomous isn't human-free.
09:45And I think it ties into what the panel is saying.
09:48And what does that mean?
09:50From a human-free perspective, it means that we have to take accountability.
09:55Right?
09:56So, AI needs adult supervision.
09:59It's there to bring the transformation that we need, we want to see.
10:03But it's also very important for us to recognize that it needs that learning.
10:09It needs that human oversight.
10:11And when you think about, you know, you spoke about really elevating human judgment in that.
10:17It's very important.
10:18And that's what 5.0 is about.
10:21It's not about eliminating human judgment.
10:23It's about elevating it.
10:24So, my summary would be, machines are good at running the show.
10:29That's what they're built for.
10:30Let them do it.
10:31What's more important is the role that we play in ensuring that it's responsible, it's trusted,
10:37and it has what I call a human thumbprint assigned to what it does and what its purpose is.
10:45Great.
10:45Thank you, Kathy.
10:46We'll talk a little bit more about accountability later in the following rounds.
10:50Now that we've set the landscape, I'd like to turn to the playbook.
10:53Kathy, I'll stick with you for this question.
10:56So, you've led growth and transformation across many companies.
11:00Many companies know they need to adopt AI, like you said.
11:05But what separates those that successfully scale AI versus those who are stuck in pilot stage?
11:12Yeah, and it's a really great question.
11:13And everyone in every company that they're working in has a pilot running.
11:17It's happening today.
11:18So, what I would say is the tech is ready.
11:23The tech is ready.
11:24And the different question we have to answer is the way that we work is ready to change.
11:29And when you look at what is happening with AI and the scalability of AI, 80% of that is
11:36really a cultural shift for organizations.
11:38And 20% is technology.
11:41And I think what really brings that to life is how do we pilot ourselves in a different way?
11:47So, 45% of pilots today are stuck in pilot mode because they can't get out of that.
11:52So, my recommendation would be to focus on three key things.
11:56The first thing is, when we think about pilots and really trying to make that transformational chain scale in AI,
12:03the mandate must come from the top.
12:05If it comes from the CEO, it is transformational.
12:09It becomes a business priority versus what is more a pilot process.
12:14So, and we've seen this happening in real life in the market today.
12:17So, you look at inspiring CEOs like Schneider Electric.
12:22You know, he made a very bold statement about taking 100 use cases into operation.
12:28And that was born out of the fact that we can't have an intelligent system for the future built on
12:35past infrastructure.
12:37And I think when you have mindsets of CEOs that are transforming at that level, it's important.
12:42The second basis is really about competitive advantage.
12:46Whether you're in a regulated market, whether you're in an SMB, it's irrelevant.
12:51You have to think about AI as a disruptor.
12:54The one thing that AI has given us is a level playing field.
12:58Knowledge is immense.
12:59Opportunities are huge.
13:01And the one example that stands out for me is JPMorgan Chase, where they have fundamentally transformed their business,
13:09taking what is a proprietary LLM capability and embedding that into their advisors, making their advisors more productive,
13:18but ultimately better profitability for the business overall.
13:23And the last thing I would share is the game changer.
13:26And I titled it enablement.
13:28Enablement is really the culture of the business.
13:30How are we driving trust to ensure that people actually adopt what is AI capability?
13:37And the enablement focusing very much on the access.
13:40So are we giving people tooling and looking away?
13:43Or are we creating an environment for true experimentation that will then take them into what is the right cause
13:49of piloting to transformation, to scale?
13:52I think if we do all these three things, we start seeing that shift and transformation required.
13:58Thank you, Kathy.
13:59High-level mandate, competitive advantage, how to leverage AI as a disruptor, and instill trust in employees to use the
14:08process.
14:09Rudy, I want to turn to you to go one level deeper into the process itself.
14:15Could you just tell us what process intelligence is at first and tell us what is the biggest challenge in
14:23implementing that?
14:25I love this question, of course.
14:27So process intelligence, it's a relatively new technology.
14:31It has been around for something like 15 years.
14:34And think about it like an X-ray system for your business processes.
14:37Whenever something hurts, something doesn't work, we take the data from your systems and push the magic button, and instantly
14:45you see on the screen your real process with all variations, all bottlenecks, all compliance issues.
14:51And, you know, companies like EY, for example, are using our technology to get a perfect understanding in consulting and
14:58in audit.
14:58Again, like a doctor for X-ray, if it hurts, you want to know where the pain comes from, and
15:03process intelligence is the tool to show it.
15:05So, and what was the second part of the question?
15:08What's the challenge?
15:10Oh, the challenges.
15:10Okay.
15:11I would say that maybe, you know, 10 years ago, it was the technology itself.
15:16Five years ago, it was integration.
15:20Today, it's neither of these.
15:22You know, these are not the obstacles anymore.
15:25Organizations can buy the technology.
15:27We can connect to the systems, SAP, ServiceNow, Salesforce, you name it.
15:31It doesn't matter what it is, as long as the data is there.
15:35And they can easily generate insights and really see the process as it is.
15:40But the real value is really how to create, or the real challenge is how to create value out of
15:46it.
15:46Because, you know, if you break your arm and you get the best X-ray picture in the world and
15:50you are sent home, you can frame it on the wall, show it to everybody.
15:54Unfortunately, your arm is still broken, right?
15:55So, you need to do something about it.
15:57You need actions, and you need decisions.
15:59Because only if you combine all of these, you get value at the end.
16:04And, you know, the challenge really is not making people see the process, but to make them decisions and really
16:11execute on what they see.
16:13Because we are really shaking up organizations sometimes.
16:16You know, we show the naked truth, and not everybody wants to see that.
16:20Not always, of course.
16:22So, the real challenge really is, you know, to break down these organization silos, processes, they cross systems, they cross
16:30organizations.
16:31Sometimes, even if you think about it, you know, a company has partners, has customers, has suppliers.
16:37So, these processes span even across organizations.
16:41And if you get the full transparency and you see what the problem is, you need to do something about
16:46it.
16:46You know, if I step on my scale in the bathroom in the morning and I don't like the number,
16:50if I don't start moving, the number will not change.
16:53At least not for the better, of course.
16:54I think that also goes to Cathy's point of this mandate has to come from the top level, right?
17:00So, for a top level, executives to see what's going on and then to push that forward.
17:05Yeah, and, you know, it's really the transformation.
17:06So, to make the right decision, to execute the right action.
17:10So, I would say that if I think about the hierarchy of challenges, today it's really the organization, the transformation.
17:18Second, it's probably, yeah, the technology is the last, definitely.
17:24This is soft.
17:25You know, the technology is largely soft, but the transformation challenge is not.
17:32So, Julie, speaking to a little bit about challenges, how do companies, how should they balance the need for collaboration?
17:42We're talking about the partnership level that you're an expert on.
17:47How should companies balance the need for collaboration with the risk of being too dependent on another platform's technology stack
17:55or tech platforms?
17:58Yeah.
17:59So, maybe let me start off by saying I truly believe in today's world that collaboration is critical.
18:06You need that.
18:07There's no single company that's going to be able to do everything on their own.
18:12And if they are, there's a complete disadvantage because they're not focusing on their true USP and what value they
18:20could truly bring.
18:20So, collaboration in a complex world has gotten critical.
18:24But I'd like to take the opportunity to talk to the audience about the three C's that everyone has to
18:31keep in mind.
18:32So, if collaboration is key, there's three things you've got to watch, especially as you're evaluating your tech stack.
18:40Number one, do you have control, control of the data, control of the process, control of the IT stack to
18:49make sure it's on when you need it to be on?
18:52It's available where and when you need it?
18:56Understanding your influence and your control across that stack is going to be critical, irrespective of driving a collaboration model.
19:04Number two, you want to make sure that you're watching out for concentration risk.
19:11In other words, you don't want your whole stack being dependent on a single vendor.
19:16You need to make sure that you've collaborated to make sure you're not completely dependent on one single party.
19:24I think that's really important that you've managed that in the right way.
19:28And the last C, you have to make sure that you've got choice.
19:33You're not beholden to a single one for the entire pie.
19:37You can choose, pick and choose and select as you go, giving yourself more freedom also for the future.
19:45If you take collaboration as critical for the future, as key on top, but kind of keep in mind those
19:52three C's around control, choice, and concentration risk,
19:56you'll manage to get the best benefits out of driving collaboration, but doing it in a way that brings true
20:03benefit for all.
20:05That's a great point.
20:06I would actually add a fourth C to your picture.
20:09Maybe it's very related to choice.
20:10It's composability.
20:12So you really create, you know, you don't have like a silo or a monolithic system.
20:16You really have the choice to select and to exchange tools and functionality.
20:21Mix and choose.
20:22That's what I meant around concentration risk.
20:25We should never have a monolith system.
20:27It's, you're over dependent on a single vendor.
20:31So having that selectability and choosing, I like the adding one more C around my collaboration mix.
20:38It's cool.
20:39Thanks, Rudy.
20:40Thank you both.
20:41Olivia, let's bring it back to the operational reality.
20:44You mentioned a little bit about some of the solutions already in place at LaPost.
20:49How are these automation systems working and how are they creating value?
20:55First, I will start with AI.
20:58At LaPost, AI has become an invisible but essential layer of our industrial network.
21:07The objective is to make processes more reliable, faster, and more efficient.
21:14Over the past seven years, we have moved from isolated use cases to a fully operational AI stack.
21:24I would like to give you one figure.
21:27Today, every item that are processed in our network sees at least one AI during its journey.
21:36This transformation started with our core industrial applications, such as real-time address reading, address recognition in our mail sorting
21:47machines.
21:48Today, we have reached an incredible performance since our mail sorting machines are able to read 97% of the
21:58address, which is a very high performance.
22:01It's also a lever for quality and efficiency.
22:05If I come to our parcel sorting centers now, they also rely on three complementary layers.
22:15The first layer is a physical layer.
22:18We have conveyors.
22:19We have sorters.
22:21We have robotics.
22:22Then you have the data capture layer.
22:25Then we have scanners, multi-phase recognition, OCR.
22:31And the third one is the AI layer, the decision layer.
22:37We have various decision-making systems, such as warehouse management system, transport management system.
22:47And AI helps us driving routing, flow optimization, and operational decisions.
22:55AI also plays a significant role in transport.
22:59We are now able to leverage billions of data to have a more precise estimate time arrival of our trucks.
23:08This is really critical because if the truck doesn't arrive, then the product can be delivered.
23:13So this is very critical for our business.
23:15And we also have scaled AI to high-volume processes, such as custom data processing.
23:25But now we are moving to a next phase, thanks to generative AI.
23:30It's deliver value to our customers.
23:32And this is, for example, something you can see in our booth.
23:37It's we created a solution so that our customers who want to send a mailing campaign, they can make a
23:46prompt.
23:47And our solution will propose the design of all the advertising campaign to have a faster creative face.
23:59Now, I will come to robotics and autonomous vehicle, what we call a physical AI.
24:08And when we come to this, we can see two different perspectives.
24:15The first one is industrial indoor robotics.
24:20If you look at industrial indoor robotics, you can see that the level of maturity is quite high.
24:26In our sorting centers, we have automatic guided vehicles, for example, which are driverless robots, moving goods within our centers,
24:38within our sites.
24:40We also have robotic sorting and automated palletizing.
24:45They're all deployed at scale.
24:47They have proven economics, and they also have a high benefit on working conditions, because it can eliminate repetitive tasks
24:57for our operators.
24:59But we are also now collaborating with a French deep tech company, which is called Wondercraft.
25:07You can also see the robot on our booth.
25:09The robot is called Calvin.
25:11You can go and meet him in our booth.
25:14And the use case that we're now developing, it's a prototype, and maybe an industrial product, to be able to
25:24carry heavy loads and heavy parcels.
25:26So, this is going to be more and more mature.
25:31When it comes to physical AI, the priority is no longer experimentation now.
25:38It is industrialization.
25:40And above all, it was said by the other speakers, it's orchestration.
25:44It's how we combine robotics, software, and human operators very effectively.
25:53Now, if we come to the last mile innovations, it's almost the same as in my example.
25:59Indoor robotics is very mature.
26:01But if you go to last mile, then you have so many constraints.
26:05You have reglementation.
26:07You have also reality.
26:09So, this is largely experimental.
26:14So, let me conclude with one key point.
26:17Automation is not the objective.
26:20Orchestration is.
26:21And the winners in the game will be those who know precisely where technology creates value and where human intelligence
26:30remains essential.
26:32Thank you, Olivia.
26:33Now, let's move on to the main question of this panel.
26:37Europe industry's 5.0 vision.
26:40As mentioned, it's not just about productivity, as Olivia also alluded to.
26:45It's also about human centricity, resilience, and sustainability.
26:49Julie, let's start with you.
26:51Europe often sees trust, governance, and regulation as its strengths.
26:55But in this current fast-moving global AI race, some of them could create friction, right?
27:01So, how can European companies earn trust and responsible, turn trust and responsible AI into a business advantage rather than
27:11a reason for slow adoption?
27:13So, let me start with the slow adoption.
27:17And let me kind of challenge that.
27:19We just did a study all across the world around adoption of AI.
27:24It was a responsible AI study, so we specifically drilled into that question.
27:30Today, we realize from the companies that we surveyed, 75% are moving towards scale.
27:36So, moving out of POC into broad-based scale.
27:40How many do you think have included governance as a topic around trust and governance in conjunction with that adoption?
27:52Less than expected, I assume.
27:54It was less than a third.
27:5575% are moving at scale.
27:58Less than one-third are actually considering the governance and the trust component.
28:03That tells me that, actually, governance, trust, control are not slowing down adoption.
28:11They're running behind.
28:12We need to do something different around this.
28:15If we really believe that trust is essential, it needs to be baked in by design at the beginning of
28:23the project, understanding what the objectives are.
28:26I feel like we've moved to a world where people throw around the world the word trust like a sprinkling
28:34on your cake.
28:35Or you know what I think of?
28:37I think of the fantastic coffee latte that you get.
28:40You know they put the little cinnamon on top?
28:43That's kind of the trust.
28:44It's supposed to make you feel like you've got a real barista coffee, an excellent one that you can trust.
28:51But in reality, all it is is cinnamon with a shaker that they threw on top.
28:55As a last minute, I think Europe has a real chance to change this, to say that the LLL models
29:03have been built.
29:05They've invested trillions.
29:07But the value that we drive, the trust that we drive, is really going to be in the application.
29:12And if Europeans take the opportunity to really build that into Industry 5.0 by design at the beginning, considering
29:23trust, governance, and control, that means that you know where things are happening, when things are happening, and how they
29:29are happening, and take accountability for that, we can develop a real strategic advantage.
29:36Let me just say one more thing.
29:39I think this is super important.
29:41Trust by design happens in the beginning of a project, at the design phase, not at the end like my
29:47cinnamon.
29:49Don't mistake trust by design for regulation.
29:54I am not asking for loads of regulation and laws that we need to do this.
30:00I'm asking for a mindset change that's supported by qualifications and requirements, not massive, complex regulations.
30:09And so my ask to all of us in the room would be, I think as we look at Industry
30:155.0, we have a huge opportunity.
30:18All of the value is still to come in this journey, especially in the application phase.
30:23We focus on trust by design will drive more adoption, more affinity, and more demand, but it's got to be
30:31baked in and not sprinkled on.
30:34And we need to focus on making sure that the way we drive adoption is through simplification of regulation, not
30:41by adding more on top.
30:43Does that make sense?
30:44Yes.
30:45Trust has the main ingredient.
30:47A hundred percent.
30:47I'd like it to be the cream, not the cinnamon at the top.
30:52Thank you, Julie.
30:54Now, Rudy, this is a very important question.
30:56So if competitors can automate faster, they can cut costs and make decisions with less human intervention, how can a
31:05human-centered model still compete?
31:11I think, you know, this question is based on a slightly almost false choice.
31:17It's not H-I or A-I, it's the combination.
31:21And if you look into the history of Europe, you know, we have decades of experience, really, in manufacturing, engineering,
31:29logistics, complex business operations.
31:33And to make A-I work, there's one thing we didn't discuss or didn't mention yet, it's context.
31:39You know, sometimes I think if you just take A-I and plug it into your operations, it's like you
31:45hire the best graduate from the best university.
31:50You know, this person is extremely smart.
31:53They know everything.
31:54Or she or he knows everything.
31:56But would you make this person CEO of your company the first day?
32:00I wouldn't.
32:01You know, because this person has no clue how the business runs.
32:05He doesn't even know where the coffee machine is.
32:08You know, that's really a challenge.
32:11And without context, A-I is a great generalist.
32:14So you can say, hey, I see a spike in demand.
32:17Should we reorder?
32:18Oh, yes, of course.
32:19Go ahead.
32:20I mean, you can ask a 10-year-old, if everybody wants ice cream, should I make more ice cream?
32:24Yes, of course.
32:25But with AI, you know, really provided with all the important context, you get a slightly different answer.
32:32You get an answer from a specialist.
32:33You know, maybe in EU we see a spike in demand for SKU 102, whatever it is, 75%.
32:41The lead time in our ERP system indicates 10 days, but 90% of the time in the past, it
32:48was 20 days when ordered or when shipped by train.
32:52For this spike, we recommend, you know, to approve flight delivery or delivery by plane.
33:00And this is the answer and the decisions of what we really need from AI, right?
33:04It's not just general answers.
33:06It's very specific for that.
33:08But AI decisions, AI actions needs to be grounded in the real processes, full understanding what's really going on, trained
33:15on everything from the past, and execute the right actions.
33:19And, of course, there's a lot of orchestration and automation.
33:22But really, you know, for me, orchestration is like the conductor in front of an orchestra.
33:28So we have all the different players.
33:29We have API systems.
33:31We have ERP systems.
33:32We have RPA agents, AI agents, humans in the loop.
33:35And in order to create the most resilient, efficient, and safe process, you need to orchestrate all of them.
33:42If the orchestrator is missing, the intelligent orchestrator is missing, it's more like the tuning of instruments in front, in
33:49the beginning of an opera, of an orchestra, or a symphony.
33:51It's musical noise, but it's not the masterpiece, you know, the audience is expecting in the room.
33:56And, of course, AI plays an important role in that.
33:59But the oversight, you know, understanding what happened in the past, understanding where we are today, and predicting the future
34:05because we know the patterns from the past, that's the real power of process intelligence.
34:11I like the analogies that's going on, like coffee and concerts.
34:15Kathy, so we're bringing...
34:17A lot of C's.
34:17Yes, more C's.
34:19Kathy, we're bringing it to the accountability question that you're an expert on.
34:24So when an autonomous system makes a bad operational decision, you know, or a wrong one, who should be accountable
34:31and what framework should that be in place?
34:34Yeah, I mean, that's a big question.
34:36But just before I answer that, I mean, fascinating oversight from the panel.
34:39I think, Julie, I loved what you said about baking in, because that's the reality of accountability and how we
34:46truly sponsor AI to be responsible moving forward.
34:49I think when you think about that question in businesses, it's about the courage to rewrite your organization around AI
34:58and the leadership to carry your people on that journey.
35:01I think if that is a cornerstone we can really help drive what is the next phase of AI, it
35:07would be impressive.
35:08I think when we think about accountability, such a big word, and autonomous is so dynamic.
35:16The two don't go hand in hand fundamentally.
35:18And the big challenge in all of this is how do we keep AI responsible?
35:24How do we keep AI scaling?
35:26How do we keep agentic AI evolving with all of the right disciplines around it?
35:33And there's three things.
35:35I think when a business thinks about operationalizing AI, you've made a decision.
35:40You've made a decision to make AI core to your processes.
35:44So you have to take the responsibility at an organization level to look after it, to manage it in the
35:52right way.
35:53And it's just as simple as a drug in pharmaceuticals.
35:56They produce the drugs.
35:58The company takes ownership for that.
36:00So for me, it starts with the organization.
36:02The second bit is around executive leadership accountability.
36:07And I think this is hugely important.
36:09We spoke about CEO mandate from the top, transforming change, hugely exciting.
36:14But when we don't leverage the value of what is true accountability at an exec level, we miss the challenge
36:21to bake it in, to your point.
36:24So I think having that focus on executive leadership is critical.
36:30And the last bit is the builders, right?
36:33The developers.
36:34They are phenomenal in the skill to be able to bring tech.
36:39And AI is, you know, you can go into so much technical terms and jargons.
36:43But what they do every day to really harness engineering behind AI is impressive.
36:49And while we have to flourish that talent, we also have to give them the responsibility around morals, right?
36:56AI doesn't carry consequence.
36:59It doesn't understand morals.
37:01So for a developer, they have a duty of care, but they also have the right to do the right
37:08thing by developing, ensuring things are tested properly, and there are guardrails to protect AI as it goes through that
37:15process.
37:16So fundamentally, it's shared accountability.
37:19Not one person is ultimately accountable.
37:21We are all hugely excited about the potential, and we see the value.
37:26The risks are enormous, and we have to be really responsible across that shared, what I call accountability.
37:33And the last thing I would say, the EU Act, which is coming on the 2nd of August, mandates that
37:40accountability and the transparency that we all now have to sign up for is real.
37:46And to my point, I'm excited.
37:49I think AI is giving us something that we've never had before.
37:54And for us as humans, for us is the ability to make that difference.
37:58We should embrace it, and we should be even more impactful in how we govern it, because the governance has
38:05to be more deliberate.
38:06Thank you, Kathy.
38:07And Olivia, let's end with you.
38:10Bringing back to how it affects workers, right?
38:13So automation directly affects their daily routines, and even sometimes their sense of job security.
38:19So very briefly, could you tell us what kind of skills will be most important for frontline workers?
38:26Actually, La Poste roughly lost two-thirds of its mail volumes between 2008 and 2025.
38:35We've been from 15 billion letters to less than five.
38:40This is really a very sharp disruption.
38:44So one answer would have been, well, let's increase automation and let's focus on the adaptation of our operating model.
38:52But what we did, we asked another question.
38:56What does France have in every village, in every street, in every morning that no tech company has, that no
39:06drone, no algorithm can replace?
39:09So the answer was a trusted human being with a local relationship, and we built a business model around that.
39:19Actually, we launched new services on the existing human delivery networks, means we used an existing asset, an infrastructure network,
39:32to transform it into a physical network.
39:34So every women, post-women, went to a male carrier into a multi-service local agent.
39:43So automation was the means, but human proximity was the strategy.
39:50Let me give you some examples.
39:52Last year, our postmen and women delivered 15 million meals to elderly people.
39:58It's a very easy solution.
40:00The family pays a subscription, and municipality co-finance it.
40:05It's apparitionally impossible to automate.
40:09It requires human presence, and it also requires trust.
40:13I did one route with a postman, and it's incredible how much the postman is weighted and how much the
40:21elderly people are happy to see him.
40:23It's sometimes the only person they see in the whole day.
40:28So we also created another network, which is a B2B solution to deliver logistics and to help brands.
40:37There are several vertical markets.
40:40We have, for example, retail shops.
40:43We have spare parts, automotive spare parts.
40:47We also have cosmetics or health, and we're delivering products to the retail shop.
40:53So even though it's similar, we trained our postal workers into these new jobs because we invested a lot in
41:05training,
41:05and so to compete on the market with other companies.
41:14This is really one more example of how we built a physical network, training our postmen, adapting their jobs,
41:24even though it's similar, adapting them to this new logistics market.
41:29Thank you, Olivia.
41:31Thank you so much, everybody.
41:32This is a great panel.
41:34Now, I think what we've heard today is the autonomous future is not simply about letting machines take over every
41:40decision.
41:41In fact, it's quite the opposite.
41:43Olivia, thank you for the real use case examples on hybrids.
41:47Julie reminded us the three Cs, control, concentration, risk, and choice, and no cinnamon.
41:54And Rudy highlighted the importance of process intelligence and how to leverage that.
41:58And Kathy, of course, demonstrated the key to accountability and baking it in the brighter beginning.
42:04Now, the real test for Industry 5.0 is whether human centricity can become a real core competitive advantage.
42:12So whether you're a startup building the next generation of industrial tech,
42:16a corporate leader looking to transform your operations,
42:18or investors looking at the next stage of innovation,
42:22I hope this panel has been helpful to you.
42:25Thank you so much for joining us, and enjoy the rest of VivaTech.
Comments