The Knowledge of Your Company at Your Fingertips: A Conversation with Udo Würtz, CDO of Fujitsu’s Platform Business
Key players insights
With a career spanning over two decades in the IT industry, Udo Würtz, Chief Data Officer of Fujitsu’s Platform Business, has been pivotal in driving Data-Driven Transformation for Fujitsu’s customers and partners. His expertise in AI and data management has significantly contributed to shaping Fujitsu's AI strategy and offerings. In this interview, Udo shares insights into the transformative power of Generative AI and the unique benefits of “Fujitsu's Private GPT Solution”, particularly in the context of the global AI landscape. Now, let's dive into the conversation with Udo.
What are the key benefits of implementing a private GPT for an enterprise, particularly in the context of the global AI landscape?
The benefits of private GPT solutions are crystal clear. They ensure that data remains within your data center, providing a level of control not possible with public cloud services. From a cost perspective, for example, the flexibility to choose open-source models translates to lower costs and greater transparency regarding data details. Scalability becomes straightforward. Specialized models tailored to specific data and usage scenarios distinguish private GPTs from generic models found in public services. This specialization empowers businesses to train models according to their needs, ensuring optimal service delivery.
What particular use cases are you already working on and how do you ensure data sovereignty in those projects?
Generative AI or Large Language Models (LLM) find applications in various sectors. When discussing generative AI and LLM, it's essential to distinguish between various types of offerings. On one hand, there are standard offerings where customers can train their own documents, for example. This trend is gaining momentum, with increasing requests, especially for Proof of Concepts (POCs) and projects. In manufacturing and automotive, customers use it, for example, to train service manuals. Details like the tire pressure of a specific vehicle or its weight-carrying capacity. Similarly, in the transportation sector, customers seek to provide comprehensive information for transporting goods, considering factors such as requirements, allowances, and regulations in different countries. Another category of customers, in the field of tax consulting, utilizes generative AI to offer quick access to information and faster responses to queries supplemented by links to official documents. In more complex LLM, such as software development, sharing code snippets or creating cloud services, private GPTs prove invaluable. Customers may also provide cloud services for their clients, where generative AI aids in issue resolution without the need for a hotline, such as unlocking an account after a mistyped password. Data sovereignty is assured by operating within your own data center, within a familiar environment and ensuring control, GDPR compliance, and transparency.
What is necessary to build a specific business case and what is Fujitsu’s role in this context?
At Fujitsu, we embrace the Human Centric Experience Design (HXD) approach, co-creating with customers and partners to identify ideas and targets. Interestingly, generative AI requires minimal business case development as it appears to act like a human. And that is exactly why humans themselves have so many ideas for potential use cases. For example, you only need to think about where employees are missing in the company, where routines cost valuable working time, which customer service could be improved, etc. The focus shifts to automating services that should behave like humans, opening up numerous use cases. Private GPTs, as we recently discussed at a CIONET event, have already spawned hundreds of use cases, showcasing their versatility. One company already had 180 use cases, and another had more than 700. This clearly illustrates that when optimizing use cases, even simple aspects like your website can be reimagined. Rather than investing significant efforts and funds in making your website elaborate and continuously updating documents, theoretically, you only need a chat-based website where users can pose questions and receive AI-generated answers. This ensures the availability of the latest information. Another example are call centers. Despite a shortage of staff and skills, not all hires are super-trained from the outset. With generative AI, employees can immediately access information when receiving a call, guiding them on how to handle the conversation effectively.
It's a big decrease in complexity in term of usability. What does this mean for the individual employee?
It's truly the knowledge of your company at your fingertips. It makes knowledge accessible for everyone and improves the decision-making process of the single employee by its high usability. In my opinion, AI isn't a threat to jobs; instead, it streamlines tasks, especially those that are repetitive. With a shortage of skilled staff in many countries, automating such tasks becomes essential. Generative AI aids in making companies more attractive, freeing up employees to focus on creativity and self-improvement.
Can you discuss the role of Fujitsu’s ecosystem partners in the deployment of private GPTs?
Fujitsu has strategically developed an ecosystem of partners as an integral component of our Data- Driven Transformation Strategy (DDTS) over the past three years. Even with a workforce of over 120,000 employees, meeting all market demands is beyond our capacity. Addressing these demands requires a substantial amount of data coupled with the requisite knowledge to implement tailored solutions for customers.
Fujitsu, for example, works with up to a thousand data scientists who make it possible to respond to particular requirements of specific projects within literally 24 hours. To make it clear once again: Today the project idea - tomorrow the specialists! Customised according to the technical requirements. Contracts, NDAs, etc. - Everything is already prepared. This cooperation initiative enables us to tap into the specialised "superpowers" of our partners, who cover various areas such as transport, object recognition, 5G technologies and a variety of other AI-related areas. The synergy with our ecosystem partners not only accelerates the launch of generative AI projects but also ensures efficiency, for example by also integrating ready-to-use solutions to deliver innovative and customised results for our customers.
How does the Fujitsu AI Test Drive provide practical hands-on experience for businesses?
To simplify the complexities of infrastructure in the realm of AI, Fujitsu introduced the “AI Test Drive”. This is an AI test infrastructure that consists of hardware and software from leading technology companies and offers a hands-on experience. Located at Cyxtera in the UK and NTT in Frankfurt, these environments provide the flexibility to test and develop your solutions, allowing you to ascertain the optimal infrastructure size. Anchored by our ecosystem partners this platform guarantees a sturdy and effective testing environment. For instance, we have integrated SUSE container technology that we have been partnering with for over two decades now to provide our customers maximized uptime on mission-critical systems and deliver business innovation faster through the adoption of containers and AI through the combination of open-source technologies and extensive expertise in crafting business solutions. Besides, we are using Juniper networks that are a pioneer in networking technologies and continues to redefine networking for the AI and cloud era continuously. Through this user-friendly and comprehensive approach, Fujitsu aims to demystify AI infrastructure complexities, fostering a seamless and informed journey for those interested in AI landscape. This makes investments in AI infrastructure transparent and less daunting. In a practical test, so to speak, we can determine the "right" AI environment free of charge using a real use case and only then go into procurement.
How does Fujitsu’s Private GPT solution integrate with existing enterprise systems and how does it handle large-scale data processing?
Our approach is to provide our customers with a bulletproof solution that really works, combining both infrastructure and software. We know, of course, that customers have their existing environments and can theoretically implement AI in those environments, but from Fujitsu's perspective, there are several challenges for us when you're dealing with a thousand customers, each deploying different elements, infrastructures and software configurations that differ in terms of virtualisation, containerisation and other factors. Our solution and our approach therefore consist of proven infrastructure elements that are scalable, so that a customer does not necessarily have to buy the infrastructure outright but can work according to a pay-per-use model. Payment is then based, for example, on usage, the number of nodes, storage capacity, computing power, etc. This can be tailored precisely to the customer's needs. Fujitsu's pay-per-use model, called uSCALE, also enables billing according to usage. This is a scalable and financially prudent approach, where costs are based on actual usage and greatly simplify the customer's budget planning.
What is the outlook for private GPTs in the enterprise sector and how is Fujitsu positioned to contribute to this future?
The outlook for the future is very promising from my perspective. We recently hosted an AI Executive Summit where a customer shared their experience with publicly available GPT solutions, revealing a price tag in the multiple five-digit range per month, starting from a few hundred euros. This starkly illustrates the risks and costs associated with not considering a private GPT solution. With paramount control over data sovereignty, transparency, cost efficiency, and energy efficiency, private GPTs emerge as a crucial investment. Fujitsu's dedicated commitment to private GPT solutions aligns seamlessly with prevailing market dynamics and presenting customers with a dependable and cost-effective pathway for the adoption of AI. This strategic alignment positions Fujitsu at the forefront of providing innovative solutions that not only anticipate market needs but also contribute to the evolution of AI technologies sustainably and efficiently.