Jean-Michel Garcia of BNP Paribas, group chief technology officer, discusses how the bank organizes adoption of artificial intelligence across a global organization. Garcia emphasizes they draw on two decades of analytics and model development to support scalable, production-grade deployments.
Hosted by John Furrier of theCUBE and presented as part of theCUBE Research, the conversation examines the bank's AI factory approach, governance structures, developer tooling and use cases such as retail virtual assistants, coding assistants and infrastructure decisions.
Key takeaways include balancing speed and industrialization; building an AI factory and a token strategy; retaining on-premise control for sensitive workloads; enabling experimentation at the business unit level while enforcing group governance; protecting client data and avoiding indiscriminate data exposure; and adopting multiple tooling options to prevent vendor lock-in and optimize inference costs.
This discussion addresses artificial intelligence governance, operational strategy, model governance, developer experience, inference cost optimization and data protection in banking. It provides practical insights for financial services professionals and technology leaders seeking to scale AI initiatives across complex organizations.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
IBM Think 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for IBM Think 2026
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for IBM Think 2026.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
IBM Think 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to IBM Think 2026
Please sign in with LinkedIn to continue to IBM Think 2026. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Jean-Michel Garcia, BNP Paribas
Jean-Michel Garcia of BNP Paribas, group chief technology officer, discusses how the bank organizes adoption of artificial intelligence across a global organization. Garcia emphasizes they draw on two decades of analytics and model development to support scalable, production-grade deployments.
Hosted by John Furrier of theCUBE and presented as part of theCUBE Research, the conversation examines the bank's AI factory approach, governance structures, developer tooling and use cases such as retail virtual assistants, coding assistants and infrastructure decisions.
Key takeaways include balancing speed and industrialization; building an AI factory and a token strategy; retaining on-premise control for sensitive workloads; enabling experimentation at the business unit level while enforcing group governance; protecting client data and avoiding indiscriminate data exposure; and adopting multiple tooling options to prevent vendor lock-in and optimize inference costs.
This discussion addresses artificial intelligence governance, operational strategy, model governance, developer experience, inference cost optimization and data protection in banking. It provides practical insights for financial services professionals and technology leaders seeking to scale AI initiatives across complex organizations.
play_circle_outlineBNP Paribas CTO Jean‑Michel Garcia on Managing Infrastructure and Security for Europe’s Largest Retail, Investment, and Asset Management Bank
replyShare Clip
play_circle_outline1) Balancing Rapid AI Innovation with Enterprise-wide Industrialization and Governance for Customer and Employee Experiences
2) From Fast Innovation to Federated Governance: Scaling AI Across Customer and Employee Experiences
3) Scaling AI: Speed, Industrialization, and Governance for Customer and Employee Transformation
4) AI at Enterprise Scale—Balancing Speed of Innovation with Governance for Customer and Employee Impact
5) Rapid AI Transformation Meets Governance: Industrializing Innovation Across Customer and Employee Journeys
replyShare Clip
play_circle_outlineFrom Committees to Operations: Group AI Governance Drives €750M Annual Value Across 1,000 Use Cases
replyShare Clip
play_circle_outlineVirtual assistants in retail banking flagged as an early success.
>> Hello, I'm John Furrier with theCUBE. We are here at theCUBE's studios here in Boston's the IBM Studio with theCUBE here. I'm John Furrier, host of theCUBE with a great guest. Financials are a big bank in Europe. Obviously global financial services is hot. AI is changing the game. Jean-Michel Garcia of BNPP is here. Thanks for joining me. Thanks for coming on.
Jean-Michel Garcia
>> Thanks for the invitation.
John Furrier
>> Share the scope and size of your organization in terms of financial services scope. Very, very large.
Jean-Michel Garcia
>> So BNP Paribas is the largest bank in Europe. So we are doing the full fledge of businesses, so retail banking, investment banking, global market banking, asset management, insurance. So everything related to clients or customer or market.
John Furrier
>> So you have a big budget.
Jean-Michel Garcia
>> For IT, yes. It's a big budget. It's a big bank. And I'm a group CTO of BNP Paribas, so I'm managing roughly 12 or 13,000 people. And I'm in charge of the infrastructure, the application security and the production security for the world banks.
John Furrier
>> Yeah. I mean, financial service, you guys always been on the cutting edge of technology. IT's been a differentiator. AI's come in and there's opportunity. Machine learning's been around for a long, long time. Fraud detection. Gen AI has always had a high bar from a security posture standpoint. I won't say reluctant, but normal review, it's a high bar. When you look at the transformation with AI, how do you think about that in your organization? What is the mandate? How do you talk about that? Or do you say the bar is this high before we adopt anything? How do you guys take that journey?
Jean-Michel Garcia
>> Well, as you said, we've been working for models and algorithm and for a while now, almost 20 years, so it's not something new. Which came as a surprise two years ago with the impact, the scale of the transformation and the fact that in on single newt of technology, we were able to maybe transform all the kind of business we used to have, all the employee experience, all the customer experience. So nothing compared to the previous transformation or even revolution. This one is a massive one. But maybe because our experience, or thanks to our experience, we try just to keep the right posture, to adapt the right speed in term of transformation. So it's really important for our bank to ultimately think about the clients, the customer. We are here to sell services, but also to protect and to sell. So it's really important to keep that in mind. So I would say it's important to free up the innovation. So business by business, or entity, or function by function, they have to be able to play experiment and potentially put in projection first, but we have to keep in mind that in one year, two year, three years, we will have to extend to the whole perimeter of the bank and then we will have to keep the kind of industrialization aspect. So it's a kind of a balance, a trade off between speed and let's say overall transformation and .
John Furrier
>> It's very hard. I mean, I can appreciate what you do because you have to be forward thinking because you're in financial service and you need the edge. You need that transformation edge. So I guess my question is, how did you take that journey? Take us through. I mean, IBM is always talking about client zero. Jim Kavanaugh, who I interviewed, the CFO, talks about the leadership principles he has. How did you set this up? Take us through the mindset and then the execution of the transformation. By department? Was it one? Did you get a couple people together?
Jean-Michel Garcia
>> Well, the fact of it's maybe a good start, we tried to avoid the kind of digital journey we had to go through 10 years ago, and every banks and every industry we made the same mistake. So we put everything and everybody on the same table and it was a mess. It was a big, big mess.
John Furrier
>> Too many cooks in the kitchen.
Jean-Michel Garcia
>> Exactly. So this time we tried to be organized. First, we set up a large AI grievance at group level. So business by business, function by function, we tried to set up a core framework to, let's say, list all the new technology, to list the first use cases. In BNP Paribas, we have almost now 1000 use cases ongoing with a 750 million value on the yearly basis. So it's progressing well. We are well structured. Every businesses deploy its own governance, so it's a kind of . And at the exec level or operational level, there are committees where we can observe. So that's the first point.
John Furrier
>> Where was the results? Where did you see results? Where were the wins and what were the learnings?
Jean-Michel Garcia
>> Well, the learnings, we have to be consistent with the strategy and I will maybe discuss about the strategy later on. We need to very quickly connect the need and the demand with the offers, and that's why we need to build a strong AI factory capabilities inside the bank or by making some partner, external partnership. And then we can see some benefit business by business or where we see new value for the clients or new value. I'll give you an example. For example, in Europe with the retail banks, we are deploying a virtual assistant for the customer and it's going well at the right pace and it's fantastic for the clients because it's completely the new way to interact with their bank. So that's the first success. New other will come, but-
John Furrier
>> Yeah. Well, you mentioned strategy, you guys operate in a regulated industry. What was the strategy?
Jean-Michel Garcia
>> The strategy is to, let's say definitely to use AI, not to become AI, to use AI to better sell our customer and to give more value to them, to be able to also to extend the level of services in whatever projects they will adopt. But also internally, for employees, to give them more capabilities to focus on the right set of task and to, let's say, make them more efficient and more focused on what they do. And basically they will give new things to consume to the consumer and the customers.
John Furrier
>> So you are creating AI builders.
Jean-Michel Garcia
>> Yes, definitely.
John Furrier
>> In your company.
Jean-Michel Garcia
>> AI enabler and AI builders.
John Furrier
>> On the building side, we saw that the coding was a great win because you can see proof. How is that going and how's the coding numbers? Has it surpassed human coding yet? Some companies have probably, "Hey, we just have the more agents coding now than we do humans." But it's been a benefit. Can you share how you guys experienced coding and even talk about Bob if you're evaluating that?
Jean-Michel Garcia
>> Which is funny, the coding assistant went lately on the table. At the beginning, it was just a way to use LLM and the developer were using the LLM to generate codes, but it was not well, let's say, included in the tool chain activities or even on the global project. And only one year ago, it completely spread out, and the notion of skills and how you can be one developer but doing almost all the task which were done by many people before, and the explosion of the tool chain itself, I don't know in the future-
John Furrier
>> Was the reason the tooling got better or people got comfortable?
Jean-Michel Garcia
>> Both.
John Furrier
>> Both?
Jean-Michel Garcia
>> Both. I think they were ready to adopt. They were able to see a benefit. I think there were also, let's say the business seeing the value and the speed of the innovation. It's a kind of new ecosystem and it's really important for us to keep the pace in that aspect.
John Furrier
>> Andy Grove from Intel used to have an expression, "Let chaos reign, then reign in the chaos." Has there been a consequence of the coding with either agent deployments? You're evaluating Bob, which is more than a coding assistant, it's got more compliance in it. You have to think about all these things.
Jean-Michel Garcia
>> I think we are evaluating Bob because Bob is answering to a real problem. The problem is what do we do in the future regarding the partnership and the evolution of the fast evolution of the different coding model? So one year ago it was a ChatGPT. Six months ago, it was Gemini. Now it's Claude. Next week, maybe it will be ChatGPT 5.5 because this one is better and there will be another one and another one. The initial strategy of BNP Paribas was to invest on an AI factory and a token factory and to run LLM on premises, LLM as a service on the open model. One year ago for the coding assistance, it was great. Now, the last month, honestly, the open world on the model aspect didn't follow the rhythm.
John Furrier
>> So you went to get your tokens from a startup provider?
Jean-Michel Garcia
>> Yeah. So it's a balance between CapEx and OpEx for regarding the tokens.
John Furrier
>> AI factory to get free tokens costs more than the tokens.
Jean-Michel Garcia
>> Exactly. So now you need to look at the performance of the model, but performance also in term of context, which kind of volume of context do you need? And then with Mythos, you will have also to clean the code you want to generate or you have generated. So it's a new way, it's a new pipeline. There will be an upstream pipeline, downstream pipeline. But on our side, for sure we will have to keep the control, especially on premise. We have very good model will come very soon for the-
John Furrier
>> Well, the prices are dropping on some of these non-big NVIDIA systems. I mean, NVIDIA's a monster machine or cluster, whatever you want to call it, supercomputer or rack. Now you can get tokens coming out of smaller clusters. Are you seeing that kind of distributed architecture?
Jean-Michel Garcia
>> Yes, exactly. And we want to specialize inferencing against training. We don't do that much training because we are not a model builder. But we need to specialize. And I would say with the volume of data, you want to ingest in a model at some point, one time a year or two time a year, you will have maybe to train your model also, but we want to do-
John Furrier
>> Reinforced learning, that's different.
Jean-Michel Garcia
>> That's different. And that's more a kind of data scientist approach. But from the inference and the token price, we want to reach a neutral cost. So if I need to go on something else, something different than the GPUs, I will go something different. I just want to reach .
John Furrier
>> You want the biggest context window possible and the best reasoning capabilities. How are you injecting intelligence into your system but with your operations? You mentioned the coding. I've seen departments, "I love Cursor, I love this." So have you seen some of these developer builder behaviors where they get comfortable with their tools and fight for each other, against each other, or is it more democratic?
Jean-Michel Garcia
>> Especially for the developers, sometimes they can be difficult to follow and to mix with the standard framework. But I will say this time, we wanted to organize a large study where they will be able to test a direct channel SaaS approach with the different player, but also Bob, there's also Cursor, also GitHub from Microsoft. And we will select ... I'm pretty sure that ultimately we will have several decision to make. We will make several choice because nobody will agree on one tool. But maybe that's good because at least in term of capability to jump from one technology to another, or from a pure vendor locking aspect, maybe it's better.
John Furrier
>> Well, you also, I would imagine a proponent of choice because you want to have the ability to go out for tokens if they're cheaper, get them inside, model choice, probably similar philosophy, go with the best model and decouple that from the system. Is that something that resonates with you?
Jean-Michel Garcia
>> Yeah, clearly. And then it took us almost 20 years to converge the different tool chain. So I cannot ask them to converge it to one single ID/LLM or agentic in one year.
John Furrier
>> Well, you guys are very disciplined. Also, again, you're regulated, but also you're moving very, very fast. What has been the big operational approach? We saw with cloud native, DevOps, DevSecOps was a great movement, created a lot of agility. What's the equivalent version for AI? Is there a philosophy or cultural mindset that's different than cloud native? Because it seems more accelerated, faster.
Jean-Michel Garcia
>> In my opinion, what is very difficult this time is for the cloud transformation and the decision or the strategy we made collectively, all the banking industry, but every other company to go on the cloud to define the journey, whether or not, or which portion of the information system you will move, blah, blah. It was based on the fact that we had the last 20 years experience to run our infrastructure and we had the last 40 years of experience to run IT. So we had some big professional, we have a culture, we had a history. Here, we have to make decision without any past. That's totally new. So when we take the decision, for example, to say, "Okay, I'm going to build ... My strategy is to build first an AI factory and I will buy GPUs, I will run my model, I will define data model hub. I will clean the data, blah, blah." It's totally new.
John Furrier
>> You're a frontier practitioner.
Jean-Michel Garcia
>> Yes, so it's-
John Furrier
>> You're plowing the fields.
Jean-Michel Garcia
>> So that's why it's complex.
John Furrier
>> With a team that has so much experience doing the hard stuff for 20 years.
Jean-Michel Garcia
>> So you must agree, you must accept the fact that potentially you will make a decision, but you will adapt slightly your strategy because the context has changed, because the new technology and because at that time you didn't know completely.
John Furrier
>> So is it fun?
Jean-Michel Garcia
>> Yeah. It's-
John Furrier
>> It's a good time to be an engineer right now, don't you think?
Jean-Michel Garcia
>> .
John Furrier
>> I mean, I always say, wish I was 25 again, given what I know now, I mean, it really is innovative environment.
Jean-Michel Garcia
>> It started within bank by doing telecom, and it was a good period for telecom.
John Furrier
>> It was a good growth year.
Jean-Michel Garcia
>> And honestly, here it's a very good one.
John Furrier
>> Yeah. Well, it must be really motivating to have that exciting road ahead. Jean-Michel, thank you for coming on theCUBE. Final question for people that are in a heavily regulated environment like the finances or other? They have a background, they have data, they've done a lot of tagging, they've done a lot of work on the data. What's your advice for the hybrid AI era for folks in highly regulated industries?
Jean-Michel Garcia
>> Don't play with your data, don't underestimate the risk of spreading everything outside. Even you need to do things with, let's say dedicated partnership outside the bank, but clients is most important.
John Furrier
>> As group CTO, I have to ask the personal question. Hope you don't mind. Are you coding again? Did you get back in the game?
Jean-Michel Garcia
>> No. No .
John Furrier
>> We're hearing a lot of execs coming back in. Thanks for coming. I really appreciate that.
Jean-Michel Garcia
>> Thank you very much.
John Furrier
>> I'm John Furrier with theCUBE. We are here at IBM Think in Boston. Thanks for watching.
>> Hello, I'm John Furrier with theCUBE. We are here at theCUBE's studios here in Boston's the IBM Studio with theCUBE here. I'm John Furrier, host of theCUBE with a great guest. Financials are a big bank in Europe. Obviously global financial services is hot. AI is changing the game. Jean-Michel Garcia of BNPP is here. Thanks for joining me. Thanks for coming on.
Jean-Michel Garcia
>> Thanks for the invitation.
John Furrier
>> Share the scope and size of your organization in terms of financial services scope. Very, very large.
Jean-Michel Garcia
>> So BNP Paribas is the largest bank in Europe. So we are doing the full fledge of businesses, so retail banking, investment banking, global market banking, asset management, insurance. So everything related to clients or customer or market.
John Furrier
>> So you have a big budget.
Jean-Michel Garcia
>> For IT, yes. It's a big budget. It's a big bank. And I'm a group CTO of BNP Paribas, so I'm managing roughly 12 or 13,000 people. And I'm in charge of the infrastructure, the application security and the production security for the world banks.
John Furrier
>> Yeah. I mean, financial service, you guys always been on the cutting edge of technology. IT's been a differentiator. AI's come in and there's opportunity. Machine learning's been around for a long, long time. Fraud detection. Gen AI has always had a high bar from a security posture standpoint. I won't say reluctant, but normal review, it's a high bar. When you look at the transformation with AI, how do you think about that in your organization? What is the mandate? How do you talk about that? Or do you say the bar is this high before we adopt anything? How do you guys take that journey?
Jean-Michel Garcia
>> Well, as you said, we've been working for models and algorithm and for a while now, almost 20 years, so it's not something new. Which came as a surprise two years ago with the impact, the scale of the transformation and the fact that in on single newt of technology, we were able to maybe transform all the kind of business we used to have, all the employee experience, all the customer experience. So nothing compared to the previous transformation or even revolution. This one is a massive one. But maybe because our experience, or thanks to our experience, we try just to keep the right posture, to adapt the right speed in term of transformation. So it's really important for our bank to ultimately think about the clients, the customer. We are here to sell services, but also to protect and to sell. So it's really important to keep that in mind. So I would say it's important to free up the innovation. So business by business, or entity, or function by function, they have to be able to play experiment and potentially put in projection first, but we have to keep in mind that in one year, two year, three years, we will have to extend to the whole perimeter of the bank and then we will have to keep the kind of industrialization aspect. So it's a kind of a balance, a trade off between speed and let's say overall transformation and .
John Furrier
>> It's very hard. I mean, I can appreciate what you do because you have to be forward thinking because you're in financial service and you need the edge. You need that transformation edge. So I guess my question is, how did you take that journey? Take us through. I mean, IBM is always talking about client zero. Jim Kavanaugh, who I interviewed, the CFO, talks about the leadership principles he has. How did you set this up? Take us through the mindset and then the execution of the transformation. By department? Was it one? Did you get a couple people together?
Jean-Michel Garcia
>> Well, the fact of it's maybe a good start, we tried to avoid the kind of digital journey we had to go through 10 years ago, and every banks and every industry we made the same mistake. So we put everything and everybody on the same table and it was a mess. It was a big, big mess.
John Furrier
>> Too many cooks in the kitchen.
Jean-Michel Garcia
>> Exactly. So this time we tried to be organized. First, we set up a large AI grievance at group level. So business by business, function by function, we tried to set up a core framework to, let's say, list all the new technology, to list the first use cases. In BNP Paribas, we have almost now 1000 use cases ongoing with a 750 million value on the yearly basis. So it's progressing well. We are well structured. Every businesses deploy its own governance, so it's a kind of . And at the exec level or operational level, there are committees where we can observe. So that's the first point.
John Furrier
>> Where was the results? Where did you see results? Where were the wins and what were the learnings?
Jean-Michel Garcia
>> Well, the learnings, we have to be consistent with the strategy and I will maybe discuss about the strategy later on. We need to very quickly connect the need and the demand with the offers, and that's why we need to build a strong AI factory capabilities inside the bank or by making some partner, external partnership. And then we can see some benefit business by business or where we see new value for the clients or new value. I'll give you an example. For example, in Europe with the retail banks, we are deploying a virtual assistant for the customer and it's going well at the right pace and it's fantastic for the clients because it's completely the new way to interact with their bank. So that's the first success. New other will come, but-
John Furrier
>> Yeah. Well, you mentioned strategy, you guys operate in a regulated industry. What was the strategy?
Jean-Michel Garcia
>> The strategy is to, let's say definitely to use AI, not to become AI, to use AI to better sell our customer and to give more value to them, to be able to also to extend the level of services in whatever projects they will adopt. But also internally, for employees, to give them more capabilities to focus on the right set of task and to, let's say, make them more efficient and more focused on what they do. And basically they will give new things to consume to the consumer and the customers.
John Furrier
>> So you are creating AI builders.
Jean-Michel Garcia
>> Yes, definitely.
John Furrier
>> In your company.
Jean-Michel Garcia
>> AI enabler and AI builders.
John Furrier
>> On the building side, we saw that the coding was a great win because you can see proof. How is that going and how's the coding numbers? Has it surpassed human coding yet? Some companies have probably, "Hey, we just have the more agents coding now than we do humans." But it's been a benefit. Can you share how you guys experienced coding and even talk about Bob if you're evaluating that?
Jean-Michel Garcia
>> Which is funny, the coding assistant went lately on the table. At the beginning, it was just a way to use LLM and the developer were using the LLM to generate codes, but it was not well, let's say, included in the tool chain activities or even on the global project. And only one year ago, it completely spread out, and the notion of skills and how you can be one developer but doing almost all the task which were done by many people before, and the explosion of the tool chain itself, I don't know in the future-
John Furrier
>> Was the reason the tooling got better or people got comfortable?
Jean-Michel Garcia
>> Both.
John Furrier
>> Both?
Jean-Michel Garcia
>> Both. I think they were ready to adopt. They were able to see a benefit. I think there were also, let's say the business seeing the value and the speed of the innovation. It's a kind of new ecosystem and it's really important for us to keep the pace in that aspect.
John Furrier
>> Andy Grove from Intel used to have an expression, "Let chaos reign, then reign in the chaos." Has there been a consequence of the coding with either agent deployments? You're evaluating Bob, which is more than a coding assistant, it's got more compliance in it. You have to think about all these things.
Jean-Michel Garcia
>> I think we are evaluating Bob because Bob is answering to a real problem. The problem is what do we do in the future regarding the partnership and the evolution of the fast evolution of the different coding model? So one year ago it was a ChatGPT. Six months ago, it was Gemini. Now it's Claude. Next week, maybe it will be ChatGPT 5.5 because this one is better and there will be another one and another one. The initial strategy of BNP Paribas was to invest on an AI factory and a token factory and to run LLM on premises, LLM as a service on the open model. One year ago for the coding assistance, it was great. Now, the last month, honestly, the open world on the model aspect didn't follow the rhythm.
John Furrier
>> So you went to get your tokens from a startup provider?
Jean-Michel Garcia
>> Yeah. So it's a balance between CapEx and OpEx for regarding the tokens.
John Furrier
>> AI factory to get free tokens costs more than the tokens.
Jean-Michel Garcia
>> Exactly. So now you need to look at the performance of the model, but performance also in term of context, which kind of volume of context do you need? And then with Mythos, you will have also to clean the code you want to generate or you have generated. So it's a new way, it's a new pipeline. There will be an upstream pipeline, downstream pipeline. But on our side, for sure we will have to keep the control, especially on premise. We have very good model will come very soon for the-
John Furrier
>> Well, the prices are dropping on some of these non-big NVIDIA systems. I mean, NVIDIA's a monster machine or cluster, whatever you want to call it, supercomputer or rack. Now you can get tokens coming out of smaller clusters. Are you seeing that kind of distributed architecture?
Jean-Michel Garcia
>> Yes, exactly. And we want to specialize inferencing against training. We don't do that much training because we are not a model builder. But we need to specialize. And I would say with the volume of data, you want to ingest in a model at some point, one time a year or two time a year, you will have maybe to train your model also, but we want to do-
John Furrier
>> Reinforced learning, that's different.
Jean-Michel Garcia
>> That's different. And that's more a kind of data scientist approach. But from the inference and the token price, we want to reach a neutral cost. So if I need to go on something else, something different than the GPUs, I will go something different. I just want to reach .
John Furrier
>> You want the biggest context window possible and the best reasoning capabilities. How are you injecting intelligence into your system but with your operations? You mentioned the coding. I've seen departments, "I love Cursor, I love this." So have you seen some of these developer builder behaviors where they get comfortable with their tools and fight for each other, against each other, or is it more democratic?
Jean-Michel Garcia
>> Especially for the developers, sometimes they can be difficult to follow and to mix with the standard framework. But I will say this time, we wanted to organize a large study where they will be able to test a direct channel SaaS approach with the different player, but also Bob, there's also Cursor, also GitHub from Microsoft. And we will select ... I'm pretty sure that ultimately we will have several decision to make. We will make several choice because nobody will agree on one tool. But maybe that's good because at least in term of capability to jump from one technology to another, or from a pure vendor locking aspect, maybe it's better.
John Furrier
>> Well, you also, I would imagine a proponent of choice because you want to have the ability to go out for tokens if they're cheaper, get them inside, model choice, probably similar philosophy, go with the best model and decouple that from the system. Is that something that resonates with you?
Jean-Michel Garcia
>> Yeah, clearly. And then it took us almost 20 years to converge the different tool chain. So I cannot ask them to converge it to one single ID/LLM or agentic in one year.
John Furrier
>> Well, you guys are very disciplined. Also, again, you're regulated, but also you're moving very, very fast. What has been the big operational approach? We saw with cloud native, DevOps, DevSecOps was a great movement, created a lot of agility. What's the equivalent version for AI? Is there a philosophy or cultural mindset that's different than cloud native? Because it seems more accelerated, faster.
Jean-Michel Garcia
>> In my opinion, what is very difficult this time is for the cloud transformation and the decision or the strategy we made collectively, all the banking industry, but every other company to go on the cloud to define the journey, whether or not, or which portion of the information system you will move, blah, blah. It was based on the fact that we had the last 20 years experience to run our infrastructure and we had the last 40 years of experience to run IT. So we had some big professional, we have a culture, we had a history. Here, we have to make decision without any past. That's totally new. So when we take the decision, for example, to say, "Okay, I'm going to build ... My strategy is to build first an AI factory and I will buy GPUs, I will run my model, I will define data model hub. I will clean the data, blah, blah." It's totally new.
John Furrier
>> You're a frontier practitioner.
Jean-Michel Garcia
>> Yes, so it's-
John Furrier
>> You're plowing the fields.
Jean-Michel Garcia
>> So that's why it's complex.
John Furrier
>> With a team that has so much experience doing the hard stuff for 20 years.
Jean-Michel Garcia
>> So you must agree, you must accept the fact that potentially you will make a decision, but you will adapt slightly your strategy because the context has changed, because the new technology and because at that time you didn't know completely.
John Furrier
>> So is it fun?
Jean-Michel Garcia
>> Yeah. It's-
John Furrier
>> It's a good time to be an engineer right now, don't you think?
Jean-Michel Garcia
>> .
John Furrier
>> I mean, I always say, wish I was 25 again, given what I know now, I mean, it really is innovative environment.
Jean-Michel Garcia
>> It started within bank by doing telecom, and it was a good period for telecom.
John Furrier
>> It was a good growth year.
Jean-Michel Garcia
>> And honestly, here it's a very good one.
John Furrier
>> Yeah. Well, it must be really motivating to have that exciting road ahead. Jean-Michel, thank you for coming on theCUBE. Final question for people that are in a heavily regulated environment like the finances or other? They have a background, they have data, they've done a lot of tagging, they've done a lot of work on the data. What's your advice for the hybrid AI era for folks in highly regulated industries?
Jean-Michel Garcia
>> Don't play with your data, don't underestimate the risk of spreading everything outside. Even you need to do things with, let's say dedicated partnership outside the bank, but clients is most important.
John Furrier
>> As group CTO, I have to ask the personal question. Hope you don't mind. Are you coding again? Did you get back in the game?
Jean-Michel Garcia
>> No. No .
John Furrier
>> We're hearing a lot of execs coming back in. Thanks for coming. I really appreciate that.
Jean-Michel Garcia
>> Thank you very much.
John Furrier
>> I'm John Furrier with theCUBE. We are here at IBM Think in Boston. Thanks for watching.
John Furrier