S2: E3: AI‑driven workstations: Governance, skills & security

00:00:02: Welcome to the ALSO AI podcast!

00:00:06: Discover the latest advancements in artificial intelligence through insightful discussions with industry experts.

00:00:13: Each episode will bring a fresh perspective on AI and addresses their real-world challenges, opportunities & interests you as an IT reseller might have or your customers could ask about.

00:00:26: Featuring experts from ALSO and leading Vandors we explore diverse insights on AI.

00:00:35: Hi and welcome to Aalto AI Podcast.

00:00:38: In this episode, we will be focusing on AI-driven workstations And more importantly what it really takes To make them successful in the real world.

00:00:47: AI is not longer just about experimentation of What's happening in a cloud.

00:00:53: We are also seeing AI moving closer The user into professional Workstations talk about today, this is where we will look into governance skills security and how to control everything here.

00:01:06: And also together a bit closer to what it's really critical for the performance itself.

00:01:12: My name is Henrik Gassaheve.

00:01:14: I would be your host today and i have the pleasure of having good guests with me so i'll start introducing myself.

00:01:22: maybe you can do an introduction yourself.

00:01:26: I'm Maggie Hakim, and I am the

00:01:30: channel lead for AI & Hybrid Cloud Solutions at Lenovo Europe in Middle East.

00:01:38: In other words what i do is work with our channel partners and their customers to turn interest in AI into solutions that customers can adopt run-and-scale.

00:01:49: Thank you for having me here and looking forward to our conversation.

00:01:54: That will be a good conversation.

00:01:55: thank You Maggie!

00:01:56: Jörg?

00:01:57: Can we take it away?

00:01:59: Thank you very much, Henrik.

00:02:00: Yeah my name is Jörg Roskowitz.

00:02:02: I'm with AMD since seventeen years responsible for the FIEs and solution architects in central EMEA And we are working with our partners like Lenovo Like also on the proof of concepts On the solutions For our end customers Defining The right hardware as well As to software Solution.

00:02:27: Thank you

00:02:28: for that.

00:02:30: It is can you get.

00:02:30: who's the next?

00:02:32: Yeah, so I'm Idris nice to meet You.

00:02:34: i'm So glad To be here so i work at also friends and i am The project manager for the brand Lenovo And it's a pleasure to speak about AI today.

00:02:44: so thank you for having me.

00:02:46: Welcome to all of you.

00:02:47: so let's jump out Where it really matters.

00:02:52: Why workstations for AI?

00:02:54: So, Workstations For AI is of course something that there's a topic today.

00:02:59: But workstions are the powerhouses for AI when you talk about it on local level.

00:03:04: how to work with that?

00:03:06: so to you Jörg from beginning why are the long work stations and Lenovo ThinkPad P-Series workstages with AMD this strong platform to go for.

00:03:15: That's very good question.

00:03:16: Thank You Very Much!

00:03:18: I'm A Big Fan Of High End Workstation.

00:03:22: maybe just two sentences about myself.

00:03:25: I'm a little bit paranoia, so that means i want to have the hardware in place to run my AI models and train my AI Models.

00:03:37: And with P-Four high end workstations you will be capable of doing this.

00:03:45: First off all it's coming from our latest CPU architecture The AMD Ryzen Pro nine thousand series So, which is based on four nanometer.

00:03:55: Also here the development it's always going on shrinking to process making it more efficient having more course On The platform Which are very important.

00:04:06: specific if you want to accelerate compute and Accelerating compute Is starting from the traditional workstation workloads like how To cut out to desk solid works SketchUp and a lot of other applications.

00:04:22: So I'm a big fan of AI.

00:04:23: because why?

00:04:25: Since, I would say specific about one year models becoming much more efficient.

00:04:31: so that means on standard PCIe GPUs which are built in the think station you can do a lot.

00:04:39: So you can run inference, create your own large language models.

00:04:44: You can do experiments with Claude or Hermes.

00:04:47: also here it's very interesting how fast the development is going on... ...you can create your authentic AI models in these workstations combining the heavy CPU power combined with GPU power and accelerate proof of concepts which are important before customers maybe take their solution into the data center.

00:05:13: And here at AMD, we are working extremely close with Lenovo to build a foundation as well as bridge between hardware software and driver ecosystem The very cool thing you can develop on the workstation And then later on, you can put it into the cloud to scale it for example.

00:05:36: Keep a

00:05:36: local at the powerhouse make sure everything is running fine and tested in every thing like what we need there?

00:05:43: Exactly!

00:05:44: That's good handover.

00:05:46: You want to put one element into one PC in the workstation from a service point of view.

00:05:53: so how do I deal with that With the service point of view, how to protect the IP?

00:06:00: How do reduce latency.

00:06:02: What about control over what we are doing and keeping everything

00:06:05: safe?".

00:06:07: Yeah so Jörg was mentioning starting working on workstations then moving the workload into data centers And basically from what you're seeing in this field many organisations are struggling with these first steps on how turn an AI proof-of concept into a real project.

00:06:29: That's very accurate statement, and this is exactly the gap that we in Lenovo are trying to solve with our services portfolio on our hybrid AI advantage.

00:06:40: So what we're offering to customers Is infact a fully validated approach combining three elements.

00:06:49: so I see it as one plus formula.

00:06:53: The first element is to have the right hybrid infrastructure and that's where we offer our best-in-class portfolio with AT plus AI platforms.

00:07:05: The second element, it has proven AI use cases can cut deployment time for customers and help them achieve return of investment quickly.

00:07:17: And third element are professional services from ideation, when I have an idea.

00:07:25: I want to do something with AI too actual implementation.

00:07:30: so if we want to break it down two step by step approach We always start where the business value.

00:07:36: It's not which hardware should i use but what is that challenge and trying to solve?

00:07:42: yeah for example

00:07:43: starting with a problem

00:07:45: Exactly.

00:07:46: For example, my operation is too slow or do maybe I want to improve the speed and equality of my customer service?

00:07:54: Or Do i Want To Reduce Waste In My Processes?

00:07:57: So focusing on that challenge Is The First Step And This Is Where Our AI Discovery Workshops Come Into Place.

00:08:04: When We Focus On Form Pillars Security People How Much People Are Willing To Adopt AI Technology the hardware and technology itself, and processes.

00:08:16: So this is first step.

00:08:18: once it's identified.

00:08:20: we need to move to validation And Validation needs be fast.

00:08:24: either We move on or we kill It.

00:08:28: Here have a big Accelerator with our Lenovo AI Library.

00:08:33: Basically you can think about it as a catalog A catalogue of ready-to-customize AI use cases and AI agents that are already proven in the real world.

00:08:45: So customers are not starting from scratch, they're leveraging AI models which basically speeds everything up.

00:08:55: The approval

00:08:56: function

00:08:56: where you can just adapt and plug-and-play take a problem then add on to make solutions for it?

00:09:04: Yeah something like this.

00:09:06: And basically, once a pilot shows promise then we need to make sure that you have all elements in place.

00:09:11: To make it production ready and this is where many companies stumbled.

00:09:16: so That's why It's important to define your hybrid AI game Deciding what transware may be keeping some workloads local on the workstations for privacy and speed and extending others to data center or to cloud.

00:09:32: Sorry my voice is going for broader action.

00:09:36: That's natural when you're not AI, so that is good.

00:09:41: You are talking to the real me right now?

00:09:43: Yes

00:09:44: So our deploy and scale professional services basically define the right scalable architecture built in then necessary governance To ensure ethical legal and regulatory compliance.

00:09:59: And of course data security and sovereignty.

00:10:04: start small with a read use case, prove the value and then industrialize it with good process and scale when you're ready.

00:10:13: So this is the recipe that we take from services portfolio in order to help customers with their journey.

00:10:20: And I think we have talked about start small couple of times already but even more important just to start.

00:10:26: So maybe hand over, Idris.

00:10:28: You talk so much with the customers on a daily level... ...so this really means a lot for the customer.

00:10:33: what to do and how to start?

00:10:36: And how does the customer respond when you're talking about working with workloads or AI workloads at local levels doing it in our workstations?

00:10:45: What are your discussions about?

00:10:48: First off When we talked about workstation many people think of cloud but for many organizations, local AI processing on workstation is becoming a strategic choice.

00:10:58: First off because local inference means speed and with the novel thing station powered by high performance AMD AI workloads such as design automation or video denoising can be processed instantly without latency or dependency of network connectivity.

00:11:19: And then we have privacy location and which means are essential because today running AI model locally ensure that sensitive data never leave the workstation.

00:11:31: And for industry dealing with proprietary designs, confidential major files this is a major advantage over cloud-based inference.

00:11:40: Third predictable performance in cost control play key role.

00:11:44: You know, today with local workstation organization avoid variable cloud cost and gain consistent performance which is essential for creative professional engineers.

00:11:56: So in short AR Workstation offers speed security and control making them the ideal platform for our local inference and creative workloads.

00:12:06: Great.

00:12:07: so when we then go to next topic talking about governance It is, I would like to ask you about what services or what ideas we have in order to look at that together with the customers.

00:12:21: We can have a zero trust principle on the think shield.

00:12:26: Zero trust assumes no user or devices automatically trust.

00:12:31: every access request must be verified authentified and authorized And Lenovo support this approach with the Think Shield, which is a comprehensive security framework that integrates hardware level protection identity verification and policy enforcement.

00:12:49: Together there's zero trust in the think shield.

00:12:51: ensure that our cloud runs securely.

00:12:54: we full visibility control.

00:12:58: I don't know if it's clear.

00:12:59: Yeah, it makes super clear.

00:13:01: so we can use those theory of trust principles to regulate the environments.

00:13:08: And when you continue about a security part with AMD then your firmware protection is something that's super critical as well.

00:13:17: how do you control that?

00:13:19: Good question, actually.

00:13:20: So we are doing a lot of steps with a lot features working both the hardware vendors and platform vendors like Lenovo as well as software vendors.

00:13:32: to describe security mechanisms It's kind different layer project.

00:13:38: So you have the hardware layer where you make data secure.

00:13:44: For example, we have the AMD memory guard feature.

00:13:47: so that's a feature with everything which will be stored in the memory will be scrambled.

00:13:54: if an hacker gets access to the data then all those data will be scrambled and they can't use them.

00:14:03: They also cannot replicate their data into its original statement which is an important function.

00:14:10: Another function will be the AMD shadow stack, and this I like a lot.

00:14:14: so that means it's a copy of application running on the workstation.

00:14:20: So if somebody gets access to the system and tries to manipulate code then they'll be compared.

00:14:28: Once these features detects there are abnormalities in their applications It blocks them completely And this will allow the highest possible security.

00:14:40: So that means on a hardware level, we call it A&D Secure Processor which is based on PSP Platform Security Processor Which by the way Is an open standard So also built in by other manufacturers, so that it will be compatible and compliant with the regulation.

00:15:02: Then we have the software level for example specific most of these workstations will run with Microsoft.

00:15:10: We have the Pluton security which is a mix off the fifth one hundred forty three-level.

00:15:17: It was detailed on there

00:15:20: Exactly, exactly.

00:15:21: But that is a new way of the BIOS where you scramble data just to compare it with an older version of BIOS?

00:15:32: Yes!

00:15:32: Okay so this is saved in BIOS with firmware and of course these will get updated once new features are coming in.

00:15:40: So then we can always guarantee the last and most secure way to work.

00:15:46: And yeah, I'm just to complete the layers.

00:15:49: Then you have the OAN system level.

00:15:51: so those will be security features coming from Lenovo and on top would be customer data.

00:15:58: with that kind of security mechanism You can make sure your data really stays at workstation.

00:16:08: There is no one hundred percent of guarantee that people will get access, but with these additional enhancement and mechanisms we are making sure to gather with Lenovo.

00:16:19: That the customer data will stay at the server at the workstation And it would be secured.

00:16:27: so when you have everything in place on everything stored In a right way Everything's secure there then we can maybe look into how to do the skills gap that we have.

00:16:37: So, How Can We Enable The Customers To Look Into That?

00:16:40: And for that, we have Lenovo Discover and we had a fast start within our service portfolio... ...and this is something where it could bridge the gaps from where to start or help customers go there.

00:16:52: Do you have some details about that that we can talk about?

00:16:57: Yes!

00:16:59: The first question I did mention, these steps that we take and they are exactly related to the AI Discover and the AI Fast Start.

00:17:09: These services that we offer for customers in order help them within their AI journey.

00:17:17: It's basically a full life cycle of service starting from advisory and discovery.

00:17:25: because from what we see, most of the CIOs are facing starting dilemma which is like which team's case I start for.

00:17:33: Continuing to AI fast-start which basically in ninety days a proof of concept where you can use customers with data in order show their value very quickly.

00:17:47: We're flexible where to run this proof of concept.

00:17:51: So it can run on the environment after customer, for example in a workstation or in data center but also works with cloud and our labs.

00:18:03: Do you visit customers one-to-one?

00:18:06: Or how do you support them that way?

00:18:10: That's good question.

00:18:12: We have AI solution architects pre-sales teams.

00:18:17: Basically, we have the AI Center of Excellence in Europe and Middle East that supports customers.

00:18:23: so we start with talking to our customer about understanding their challenges based on what they design for the right proof concept.

00:18:34: it can happen remotely or face-to-face depending upon a customer's preferences.

00:18:44: I think it's super relevant when normally, we talk about services.

00:18:47: We talked about how to do extended warranty?

00:18:50: How to do things like a service where you have the need for that but this is really a service that are different.

00:18:56: When you come to Lenovo Services This is one-to-one Service Where You Support The Customer In order To Get Enabled Into The AI Journey And Then They Also Have A Best Trend That Can Help Them Looking into What Their Data Are What It Can Be Where To Start and So On.

00:19:14: So that can be

00:19:15: really important.

00:19:16: It's not only correct and it is not only renewable, so this is an ecosystem play where we work with our channel partners as well in order to offer these advisory services for the customers.

00:19:32: As I said its a win-win situation Where you bring customer what he needs on our Lenovo Hybrid Advantage which is bringing the speed, execution and expertise.

00:19:48: I can talk about that as well.

00:19:51: And services like this popping up we have not seen before.

00:19:55: therefore it's also quite important to discover what is actually possible now where to get help?

00:20:01: Where do you get help is of course super easy coming from an IT industry.

00:20:05: how we work with that Lenovo as an organization, whether it will be on Lenovo products or not are super like first moves within AI.

00:20:14: So of course that could really a good way to look at together with the customers.

00:20:19: so you asked from the technology side when we looked about modern tools how make more AI accessible for power users?

00:20:29: For workstation users maybe looking engineers designers powerhouses like rendering video equipment and so on.

00:20:39: How can AMD support in that part with modern tools?

00:20:44: So this is exactly what my team is doing, working with the end customers on the software solution or the adoption to different workstation solutions offered by Lenovo.

00:20:56: This isn't a very critical part.

00:20:58: so for example we are providing microservices as you call them.

00:21:02: those are different tools that can download from AMD web page and from GitHub making sure how to run rendering applications, so those will be detailed scripts you can install on the workstation and you could work with.

00:21:22: And just coming back to an example of what we have worked last because we all know yesterday we generated pictures with AI.

00:21:31: today We are doing movie clips right?

00:21:34: The last model that was deployed is very powerful from my point perfect and optimized for a power workstation like the ThinkStation P-Four is LTX.

00:21:47: So, LTX to point three is one hundred billion model where you can create video as well as audio material And it's really like working, generating pictures in the past.

00:21:59: You're adding a prompt having Superman sliding over the ocean and then you are not getting your picture or video clip where you can see Superman flying over the Ocean.

00:22:12: It is very impressive because from my point of view specific for this kind of business.

00:22:18: This will be next feature enablement for a lot of application, from movie entertainment to generate customized content.

00:22:29: For the end customer and this off course need all the possible hardware performance in terms of CPU as well as GPU making sure that this content can be secured get developed and optimized.

00:22:43: And This is what we are doing with our partners.

00:22:46: So we are helping them getting the software adopted specifically on AMD Solution.

00:22:53: There is one more feature, We call it day zero support.

00:22:56: so that means AI?

00:22:58: We all know nothing has been written in stone everything up into air.

00:23:03: The only thing you can tell us something new will appear to market.

00:23:09: It doesn't matter if it's new drivers, your applications frameworks models.

00:23:14: There is definitely something in you like Gemma for a couple of weeks back which we are working to provide the customers on The best possible experience

00:23:24: so they video performances where We're getting now.

00:23:27: that is also why the workstations and the powerhouses within the work stations Are getting more relevant?

00:23:32: If You have That need here So eat this to the partners When you talk to the customers about how to sell their workstation, so partners are getting trained in what way?

00:23:45: What kind of needs is it that they want to listen about.

00:23:49: So some resellers might be looking more into hardware and also combined with software.

00:23:55: Is there something where market can easily adapt or training being done

00:24:02: So for organization in regulated industry like finance, healthcare or the public sector.

00:24:09: The first thing that we're talking is about security and complaints because Is it non-negotiable?

00:24:17: At the foundation AMG provide hardware and firewall level security features such as a secure boot and memory encryption.

00:24:27: This protection ensure that this system starts in trust state and that data is protected even at the lowest level.

00:24:36: On top of that operating system hardening reduced attack surface by disabling unnecessary services, enforcing security configuration aligned with complaint standards.

00:24:49: it's a good point because when we talk about resellers or someone who asked about AI, essential for us is talking about the security and Lenovo adds another critical layers through in image management.

00:25:06: And life cycle services standardize secure OS image ensure consistency across the fleet faster deployment and complaints with internal and external regulation.

00:25:19: And when you combine AMD hardware security, OS hardening and Lenovo services You create a workstation platform that is secured by design compliant by default and ready for other enterprise AR workloads.

00:25:34: So now we're going to talk about having things on premise With the PC Workstation having it with the data center, we're also looking at for the cloud.

00:25:43: When you look at the workstations and sizing the right workstation for the write use how do we with AMD ensure that they have to write combination of the writes CPU or GPU NPU as well?

00:25:58: what matters here too?

00:25:59: To look into that make the best solution.

00:26:02: I think you can always create a powerhouse.

00:26:05: But also, how can you find the right things to look at when your AI models are working in the best way?

00:26:12: That's very good question Henrik and as Meggy already highlighted.

00:26:16: so we're talking about solutions And We have kind of playbook where customers choose to configure their hardware.

00:26:25: So for example on the CPU side it starts everything with four cores, six cores eight cores twelve cores so you can really specify on the needs.

00:26:36: So are your more focused on large language models?

00:26:40: Are you more focussed to generate images?

00:26:43: movies do create maybe an AI store server with a specific need for your employees, security reasons, camera detections and all these kind of things.

00:26:57: So you can configure the memory which is also big key to drive local AI models.

00:27:04: You can provision the system specific to your needs.

00:27:08: And this is where we are working together with Lenovo, so Lenovo has a complete overview for all these different customers to provide specifications, solving the customer's needs.

00:27:22: And from my point of view this is very critical.

00:27:24: similar data center because on one side you can vary fast over provision and spend too much money in investing into a too big hardware which at end don't need or do opposite.

00:27:40: so configuring may be base model.

00:27:42: then months later will find out maybe I need more power, and this is exactly what we are configuring to get the Rift Lenovo end-to-end customer.

00:27:54: Making sure that a customer will get the best possible

00:27:58: experience.".

00:27:59: So Maggie when you do use cases in workshops with customers Are also looking at where to have AI models stored?

00:28:07: Is it on a PC or cloud?

00:28:09: And how did talk about that?

00:28:13: Of

00:28:13: course, that is definitely one of the most important steps and one of... ...the most important topics we discuss with a customer.

00:28:21: And it comes down to understanding their current needs.... ....and also the future needs of the customers so that a workload can be scaled appropriately,... ..and also meet requirements for privacy and governance.

00:28:40: So is it normal that the smaller companies start in a small or do you recommend they start maybe to invest into some bigger workstation models?

00:28:50: How are their needs, very different.

00:28:52: Or can just startup super easily?

00:28:55: Yeah and AI there's.

00:28:57: no one size fits all.

00:28:59: It's important understand customers' needs.

00:29:03: so its hard say like You should Start Like That Big or Small But you need to understand.

00:29:12: Based on the needs that, that you have as a company and also get the right questions for the companies so they can find out where we're going to start and then build the case out of them?

00:29:22: And with the case to size whether it should be notebook at workstation data center or whatever way We could find in your portfolio when we had other things there.

00:29:35: York You'll also have good input here I know.

00:29:39: Yes, you mentioned it already and Maggie made the point here.

00:29:43: So if the customer is still in a kind of development phase I think your workstation is always good If money isn't there yet.

00:29:53: take T-Fortiness with Verizon AI so that you can also run these inference models.

00:29:59: You can do experiments or small proof concept on your notebook And then later when get a sense for business Then you can switch to the workstation where you scale your project or further develop your projects.

00:30:16: Just give some examples, we had recently a customer opportunity and this was with such a workstation called an AI store server.

00:30:26: The customers were coming Jörg, we want to invest more into security.

00:30:31: We have a couple of cameras in our store and we wanna analyze the video feed.

00:30:39: So what does marketing mean?

00:31:00: So analyzing the customer behavior, so which products in the store they are interested it and maybe which not.

00:31:09: And that means to improve the products in terms of positioning, color or whatever is necessary.

00:31:21: Last but not least they also mentioned OK and now let's get an AI agent added for our employees where these employees can inform about specific stuff which is happening at the store.

00:31:35: so this product available in that specific colour?

00:31:39: Yes or no?

00:31:39: And then the AI agent is verifying in the warehouse if the product still available.

00:31:45: Or some questions, how do I need to deal with a specific customer can also be for new employees and at the end we served about three different AIs use cases where our verb station is perfect tailored for.

00:32:01: So you actually run that on simple workstation A big workstation.

00:32:06: Please don't say simple workstations, those are really

00:32:10: high... When we talk workstages I know that they have a more higher level than the PC.

00:32:15: so yes of course you're right!

00:32:17: But when it comes to not talking about simple workspaces but when you talked with customers on the workstation that we are selling from also to resellers Do you face that customers are coming to the reseller where they want their biggest workstation, and say let's start here.

00:32:39: So there may be going for a big version then take them down?

00:32:43: So I think sizing is also super important.

00:32:46: so we don't overspend money but maybe spend it in the right waves.

00:32:53: It's good question because when talking about resellers The first element that we ask about Workstation is the model governance because organizations need to define policies around which ARM models are authorized, how they're updated and where it's hosted on the workstation.

00:33:16: And next comes data classification and access control.

00:33:20: not all of them should be treated in same way.

00:33:23: sensitive datasheet must be accessible to authorized users and AI workloads should respect this classification.

00:33:33: So when we talk about the reseller, not here just for sell a workstation.

00:33:39: We really think about needs of customer or other reseller And adapt offer in function.

00:33:48: are his need?

00:33:50: Okay, good.

00:33:51: So now we have more or less everything in place.

00:33:53: but we are still having a good conversation about the workstations.

00:33:58: and so with the hybrid AI strategies We had talked about that data center In one of the previous podcast will talk about cloud base with total local PCs Copilot plus PC's.

00:34:10: Now we're having did it work stations here?

00:34:13: So York from an AMD point-of-view With the powerhouses you can deliver which your hardware How do you help the customer's balance between each devices?

00:34:24: So we are looking very clearly about the customer use case, and we're helping them to have the exact mix.

00:34:32: We call it a hybrid AI so that means leveraging CPU as well as GPU at maximum value so you can get the maximum performance per dollar or performance for dollars spent on your workstation.

00:34:46: And this of course heavily depends on the Customer Use Case.

00:34:53: This can be in the education area, for example having multi-language models which will used for transcriptions and translations within European Commission.

00:35:04: We worked at a couple of school projects where you are using workstation specific helping kids to educate specific subjects like math, physics languages chemistry which from my point of view is a very perfect use case because we know the data and have it.

00:35:26: AI will give us the capability for students to learn faster more efficiently.

00:35:35: So you have shared already some good examples about how the use cases that we've been looking into together with customers.

00:35:42: I know from the NOVA site, when having those workshops and talking to their customer there are a dialogue of what they're doing... Maybe something is confidential?

00:35:52: I don't know!

00:35:52: But maybe Maggie in one of the dialogues you had with your customers or some used case that you did with the NOVA Is it something that can be shared for us as some good example to look at?

00:36:04: Sure, yeah definitely.

00:36:05: I'm going to talk about some AI use cases that we work with customers in different verticals and what's exciting is that we're seeing AI in practice and how it's changing day-to-day operations.

00:36:19: so let me give you a first example from the manufacturing world.

00:36:24: So today We are seeing very strong results on the factory floor cameras that our capturing images directly on production line and AI analyzes them locally on a workstation.

00:36:37: Basically, the system can spot defects in the surface that are hard for human eye to catch which means issues are flagged earlier.

00:36:47: There're less returns, less waste & quality stays consistent.

00:36:53: It's very focused use case automated service inspection but it delivers clear return of investment.

00:37:00: Another example that I can give is from the banking and retail as you also mentioned before.

00:37:07: So what we did, it's we delivered AI vision solutions that are running on local workstations as edge systems inside bank branches and stores.

00:37:18: one European Bank used this to analyze queue links security feeds in real time in selected locations.

00:37:27: The goal was very simple.

00:37:28: Let's improve the customer experience, let's reduce their waiting times and also increase visibility for the branch staff.

00:37:38: This is where we work jointly with a channel partner And we provided this solution across eight hundred branches.

00:37:47: So what's exciting here?

00:37:50: even though I mentioned banking as a vertical but there is no reason why this solution would not work in other verticals, for example take transportation hubs and how we have customer experience.

00:38:03: And maybe like to finish with another...

00:38:08: Can i put the question because you talked about manufacturing an image taking photos?

00:38:14: I have heard about a case that you've had, where looking at the production facilities.

00:38:19: Where the customer producing those products.

00:38:22: they were like doing quality inspection when their product has been finalized.

00:38:27: So maybe this is an example of what we are talking then monitoring during process.

00:38:32: so actually if your product is defect or not meeting criteria you are able to actually with AI, to skip and then don't process all the rest of production.

00:38:44: So basically what we did is implemented a robotic arm that can take effect parts away from the production line?

00:38:53: Yeah so even if it's an investment in the AI It will take out costs for production afterwards And by that the AI investments can be reduced by that way.

00:39:07: And you had a third one as well?

00:39:10: Yeah, and the third is basically something that can relate to small-medium businesses into every knowledge worker.

00:39:18: so we have an example where insurance company wanted to have AI driven help desk assistant So they needed support tickets coming in.

00:39:30: They need it speed up their supports.

00:39:34: So what we did is, we implemented an AI agent that pulls up relevant documentation related to the case.

00:39:42: It summarizes them and it compares with similar past cases And suggests next steps for a human support agent.

00:39:51: so its like giving every support person research assistant That's going to help him provide best advice.

00:40:01: We see similar approaches also in the legal sector where AI can review the document, it can flag errors or missing information and speed up their routine work.

00:40:13: It helps experts to focus on what really matters and spend time effectively.

00:40:19: What I wanted say is that common thread across all of these examples are same.

00:40:24: they solve a real problem They run on accessible infrastructure like workstations or edge systems and they all deliver results fast.

00:40:33: And that's what AI looks like in practice

00:40:37: today.

00:40:37: That is also super relevant when you talk about the DH part of it.

00:40:41: So I think we can only imagine that cars driving, there should be online to the cloud All the time in order to do self-driving methods not now maybe but in the future?

00:40:56: if the car will go in and stop for red light or green light, so on.

00:41:02: So all those devices that are being processed at the edge is also something super relevant to combine what's relevant with cloud data but it can take decisions very close where things really happen.

00:41:19: So what about developing the channel and win-win model that we're looking for, as a reseller engagement or customer engagement?

00:41:28: Is there something you've also looked into... When

00:41:31: I talk about Win-Win Model it basically comes down to three things.

00:41:39: First one is how fast can you move?

00:41:42: The second one is How easy is it to execute?

00:41:46: And the third one is whether you have the right expertise when it really matters.

00:41:51: Basically, this where Lenovo's strength comes in because of our broad and mature ecosystem built around it.

00:42:00: So let us break down.

00:42:02: if we take speed for example We help partners move quickly by providing shortcuts.

00:42:08: I mentioned before that AI library.

00:42:11: This is basically what we have done.

00:42:13: We've done all of the due diligence and work in order to offer our customers, and our channel partners production-ready validated AI use cases so that they do not start from scratch but can move faster from experimentation into return off investment.

00:42:34: When talking about ease of execution I guess you could say It comes down to standardization and modularity.

00:42:44: And this is why within Renovos Hybrid AI Advantage, we offer a full integrated AIS stack including the hardware level The AI models and agent management platform.

00:43:00: And what we also do, We make them available as best recipes in the form of validated designs and validated architectures.

00:43:10: So this makes it very easy for a customer to scale architecture starting small As you mentioned before growing with growth of business and workloads while staying secure and compliant.

00:43:26: And the third point is expertise, and I guess this from my point of view where Lenovo really stands out because through our AI center for excellence or global professional services teams.

00:43:39: Most importantly our strong ISV ecosystem.

00:43:43: we can support very complex AI projects you know?

00:43:46: Lenovo is a technology partner for Formula One so work in a high-performance mission critical environment.

00:43:56: And this means that customers do not need to have all of those expertise in house, we complement and augment their capabilities so they can get the work done efficiently.

00:44:11: Of course when you talk about Channel we also look at commercial part of the world.

00:44:17: So commercially partners can attach these Lenovo AI services that we discussed.

00:44:24: The AI Discover, the AI Fast Start, our management services and by doing that they unlock strong service attach opportunities which support their incentives, their rebates while reducing their delivery risk.

00:44:41: so customers are benefiting from consistent high quality AI deployments.

00:44:46: partners benefit from faster execution and better economics And we act as the service lead backbone to make AI practical and more importantly repeatable.

00:44:59: I think you touched very interesting element here because expertise is for sure something where a lot of companies, they don't know where to start so when it doesn't know how to speed up or how to make it easy.

00:45:15: that's why companies are on different level.

00:45:19: So I think we have touched a lot of things today about the hardware, about the workstation and capabilities here.

00:45:26: How to size right things but also actually just start up with dialogue with customers on how things are being done where it will be started?

00:45:34: Where do you get some help?

00:45:36: maybe invest or scale in the end so that final wording from each one for everyone can say After listening to this podcast about where you start with the workstation setup and also with services around.

00:45:55: And from a partner landscape, I will start with you Edis.

00:45:58: maybe you can start saying okay listen into this podcast.

00:46:01: Where would you as a reseller in customer Start when closing these podcasts?

00:46:08: So is that something like what you have been thinking?

00:46:15: To sum up, AI workstations are not just powerful machines.

00:46:20: They're a trust-AI platform.

00:46:22: with Lenovo ThinkStation powered by IMD organization can run an AI locally.

00:46:29: we have high performance and full data control enforce governance and guardails through zero trust and think shield And meet security compliance requirements even in the most regulated environment.

00:46:42: AI innovation doesn't have to come at the expense of security or control, and that is actually where modern workstation makes a difference.

00:46:51: That's what I can say for our seller.

00:47:02: Sure, I made the chart.

00:47:04: So AI is no longer a future concept.

00:47:08: it's already transforming how we work in a way to solve complex challenges across every industry and We are happy to be part of that solution family working with Lenovo and also

00:47:22: And so you Maggie?

00:47:23: i believe That everyone Is ready To buy The services from Lenovo In order to get some help in running this.

00:47:30: But are there anything that you would like to add for the final ending here?

00:47:34: Yeah.

00:47:35: So I would say, The best next step will be To pick one challenge That can be solved with AI and start there.

00:47:44: Prove your value quickly And if you do that You'll build momentum fast And take it from There.

00:47:51: so this is my advice.

00:47:53: One Challenge I believe This something they could either Be saving time or money to a company that might be where it starts.

00:48:03: So I think listening to this podcast has been an investment both in time and economy, so with that i will thank all the three of you for participating in this podcast here.

00:48:13: And uh...I Think You Have Done A Great Homework In All To present A Lot Of Good Things Here!

00:48:20: Let The Audience Decide Where To Go With The New AI Investments For The Future.

00:48:24: Thank you very much for joining and have

00:48:30: a good day.