The current program

The central question is: Why are many companies still unable to generate real added value from AI despite high investments – and what do they need to do differently now?

With Agentic AI, AI doesn't just trade – it actively engages. But the leap from pilot projects to scaled deployment is only possible with new governance, sovereignty, and a transformed operating model. In this exclusive session, we'll discuss how to scale AI agents without losing control – with clear governance, full sovereignty, and an operating model that productively unites humans and AI. Using best-case examples from real-world practice, you'll learn how to create genuine business value and how to measure it. For decision-makers who want to move beyond experimentation and scale their operations.

Agenda

09:00 Welcome
Larissa Holzki, Editorial Head of TECH – Handelsblatt

09:05 Launch Handelsblatt x Infosys AI Trend Report
Stefan Hezel, Head of Marketing EMEA – Infosys
Dr. Jan Kleibrink, Chief Growth Officer & Managing Director – Handelsblatt Media Group

09:20 Discussion:
Why are many companies still unable to generate real added value from AI despite high investments – and what do they need to do differently now?
Britta Daffner, Director Data & AI – Telefónica Germany
Omar Hairani, Head of Data & Artificial Intelligence – Tchibo
Sophie Vassarotti, Head of Data & AI Volkswagen AG
Dr. Frank Säuberlich, Senior Principal Data & AI Infosys Consulting

10:00
Business breakfast
If you would like to participate in the session, please contact Agyeman Baffour at A.Baffour@extern.handelsblattgroup.com.

Text translation has transformed the communication we can review before publication—documents, contracts, emails. Language translation is different: live meetings, negotiations, customer service conversations take place in real time, and mistakes can no longer be corrected before they are heard. This development changes the meaning of trust, governance, and ROI, because the translation is the conversation itself.

This is not a sales pitch. It's an open exchange among colleagues about what AI language technology and real-time language translation mean for the way global companies work, make decisions, and recruit – because for the first time, language no longer has to decide who gets a hearing.

Agentic AI promises to transform businesses by enabling autonomous systems that can plan, decide, and act with minimal human intervention. But successful adoption depends on far more than just deploying the technology; it requires robust leadership, resilient architectures, and a culture that empowers employees to collaborate with non-human decision-makers. The opportunities are significant, but the path forward won't be easy. Let's explore how organizations can move forward quickly without sacrificing trust, control, or the long-term resilience of the business.

The GenAI boom has demonstrated the capabilities of language models. The next stage brings AI into the physical world: into factories, machines, and infrastructure. Stephan Scheuer speaks with Vasi Philomin, Executive Vice President Data & AI at Siemens, about AI agents, industrial copilots, and why industrial data, digital twins, and simulation are crucial to the success of physical AI. The central question is: How can companies move from AI pilot projects to applications that deliver measurable added value on an industrial scale?

Stephan Scheuer in conversation with

Most companies react to customer problems like this: the customer reports it, the system responds. Real change happens one step earlier: the system recognizes the problem before the customer even notices it.

Robin Thi Stoinski demonstrates live on stage how this works. In 15 minutes, he builds a voice agent (live!) that acts proactively: It makes contact, understands the situation during the conversation, and integrates the solution directly into existing processes. A glimpse into the future of customer service – not theoretical, but live and tangible.

• What distinguishes a market-ready innovation from an experiment, and how can the difference be recognized?
• Where does the transition from research to industrial practice most often fail, and who bears the responsibility for this?
• Which innovations will be standard in five years, and what needs to happen to achieve that?

The Handelsblatt editorial team in discussion with:

How can Europe transform excellent research into faster, scalable products, companies, and industrial applications? The panel brings together the startup ecosystem, industry, and policymakers.

• Which AI innovations will shape the next five to ten years – and who should develop them?
• How must startups, established companies, research and politics collaborate to turn ideas into scalable applications?
• What framework conditions does Europe need to combine speed, investment power and technological sovereignty?

Larissa Holzki in the discussion with

Fewer skilled workers, higher expectations, increasing cost pressure: BARMER is industrializing AI agents to measurably relieve the burden on key service and administrative processes in the statutory health insurance system. Together with Capgemini, the company is demonstrating in the A³ project how four use cases can be transformed into a scalable transformation program: with reusable components, clear governance, and a robust data/platform foundation. The focus is on moving from isolated initiatives to end-to-end automation – with the prospect of agentizing up to 200 core processes within three years.

Pilot projects demonstrate the potential of AI agents – but how do you transform them into a stable, enterprise-wide system? In this panel, experienced leaders discuss key questions: What mistakes occur during scaling – and how can they be avoided? When does a pilot project become an "island" – and how do you recognize when it's time to scale? What data and platform foundations are truly needed to ensure scaling doesn't fail due to interface issues? And how do you maintain cost control for token and usage fees? A discussion about the leap from isolated successes to a networked infrastructure.

Luisa Bomke in the discussion with

On November 16th, Langdock will open its doors to selected guests for an exclusive Executive Business Lunch. In a relaxed atmosphere during the lunch break, you will have the opportunity to network with executives and decision-makers from various industries.

Interested in participating? Please direct your inquiries to A.Baffour@extern.handelsblattgroup.com.

Judith Dada (Co-CEO, Langdock & Senior Partner, Visionaries) will be there as host and is looking forward to the personal exchange.

On the new Start-up Stage During breaks at the Handelsblatt AI Summit, young teams will showcase the possibilities of AI for tomorrow. On both days, four selected startups will pitch their ideas in three- to five-minute presentations to decision-makers from business, technology, and innovation, followed by a Q&A session with Moritz Schönleber.

How do you turn AI expenditures into real value? In this meetup, participants will discuss how to measure the costs and benefits of AI applications – from the effort per process to ongoing operations. Which key performance indicators (KPIs) truly help with investment decisions? Feel free to share your own experiences or any open-ended questions you have about cost calculations.

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

In his keynote, Dominic offers insights into how his startup, "Leaders of AI," is evolving into a hybrid organization comprised of both humans and AI agents. Currently, they have only 10 employees but over 100 AI colleagues, each with their own personnel file and with whom they conduct regular feedback sessions. These agents support his team in becoming a self-learning organization: The AI continuously challenges existing workflows and roles, provides targeted support, and helps the team adapt dynamically to market changes. He shares best practices for AI transformation, reveals which skills are now particularly crucial, and explains why they had to introduce AIs to guide the other AI agents.

AI is making its way into football. But how can technological innovation be used without losing sight of what makes FC Bayern special? Stefan Mennerich discusses how the club is further developing its fan experience and brand in the digital age and what role AI plays in this.

Luisa Bomke in an interview with

What conditions must companies create in the next twelve months to ensure that AI does not remain stuck in individual projects, but measurably contributes to future viability?

• What three decisions must the board of directors and management make themselves now?
• What technical, organizational, and cultural prerequisites need to be established simultaneously?
• How can you tell after twelve months whether a company is truly AI-ready?

Lina Knees in the discussion with

The figures are impressive: Gartner forecasts $64 billion in AI spending by 2026 – an increase of 63 percent. But the reality is uncomfortable: Over 80 percent of companies miss their cost forecasts by more than a quarter. Over 40 percent of their AI projects could fail by 2027.

Token costs are exploding. Employees are often working more, not less. And the real efficiency gains remain enigmatic. How do companies practically control their AI budgets? Why do studies show that intensive AI use leads to more work – due to monitoring, complexity, and mental exhaustion? And how can AI become an investment rather than a cost driver?

A discussion about the costs, the hidden pitfalls, and the crucial question: How can AI expenditures generate real added value?

Stephan Scheuer in the discussion with

The public debate on AI revolves almost exclusively around models and their attributes.
But the real challenge facing companies lies one level deeper.
The question of what can give one an advantage over other companies.
The answer lies hidden in the context that lies dormant within one's own data.

To achieve this, the importance of and control over one's own data are more crucial than ever. Mistakes in these fundamentals only become apparent later, but then reverberate consistently with every subsequent model and agent. This leads to the creation of even more data silos, redundant data copies, and architectures that constantly need to be redesigned from scratch.

Markus Grau shows which data and architecture decisions really matter now so that companies will not be laggards in 2027, but rather shapers of the next AI wave.

By 2027, competitive advantage will no longer be determined by the choice of model, but by mastery of the value chain.

When AI capabilities become a commodity: Where do real competitive advantages arise: in model access or in control over your intelligence architecture? Who controls the AI's decisions—the company or the model provider? Is a proprietary foundation model really the solution?

Larissa Holzki in the discussion with

Agent in the Loop – Responsibility and Control in the Age of Autonomous Agents

The "new" colleague works around the clock, conducts independent research, prioritizes tasks, initiates processes, and accesses company systems. For many companies in the Handelsblatt AI Circle, this is already everyday life.

AI agents are increasingly leaving the protected demo environment and assuming real responsibility – within defined boundaries, but with growing scope for action. In doing so, they are not only putting productivity and efficiency to the test, but also established roles, leadership models, and responsibilities.

We are discussing:

• Who gives tasks to an AI agent?
• Who decides what they are allowed to do independently?
• Who controls their results and decisions?
• And who takes responsibility if something goes wrong?

In this AI Circle session, we won't discuss whether companies should use agents, but rather how their use works in practice. In a confidential setting, we'll share experiences with AI agents, OpenClaw, and autonomous systems, and discuss the necessary rules, control mechanisms, and new forms of collaboration.

From AI Governance to AI Factory: How Governance Enables Orchestration, Adoption, and Return on AI

Using a practical case study from a global industrial company, this masterclass demonstrates how an established AI governance system can be further developed into the control architecture of an AI factory. The focus is on the continuous cycle from clear guidelines, through portfolio and delivery orchestration, to qualified adoption and measurable results in business processes. Participants learn how, based on this foundation, steering can make sound decisions about scaling, redesigning, or halting AI initiatives.

How can AI projects be effectively prioritized and implemented with limited resources? The focus is on practical experience: Where to begin, what to develop in-house or purchase – and how to take the next step beyond the initial use case? Exchange ideas with other participants about pragmatic approaches, obstacles, and helpful decisions.

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

AI in Action!

No transferring. No waiting. Case closed. How real-time AI voice agents resolve issues independently.

„"Please hold the line..." Customer inquiries today often end in waiting loops, transfers, and media breaks. It doesn't have to be this way. In this masterclass, Jens Heilmann and Holger Hornik from msg.group demonstrate how modern AI voice agents understand calls in real time, respond contextually, and resolve issues independently – all without human intervention. Using concrete practical examples, they illustrate the underlying technological components and the limitations of current systems. This masterclass is for anyone who wants to take the next step in automating customer contact.

AI at Scale: Lessons Learned from Building Enterprise AI Platforms

The best AI model of today may not be the best-performing model of tomorrow. How can businesses scale AI while protecting long-term investment? Using a virtual try-on solution as an example, this session explores the strategic choices behind moving from custom to commercial AI models, how to balance cost and quality, and how to prepare AI platforms to adopt the next generation of models without starting over.

How much autonomy should an AI agent have – and at what points is human oversight necessary? Discuss with other participants how tasks, access rights, and approvals can be effectively structured. The focus is on practical experience and open questions: Who checks results, who intervenes, and who takes responsibility?

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

AI agents are fundamentally changing the operating model of companies. Productivity increases, automation levels grow, ROI is measured – but one crucial question often remains unanswered: How successful is the collaboration between humans and agents?

Traditional key performance indicators (KPIs) are no longer sufficient. In her keynote address, Dagmar Schuller demonstrates why new KPIs for trust, acceptance, and human-agent collaboration will determine the future success of AI initiatives. To successfully scale AI agents, companies must measure not only their performance but also the quality of their interaction with humans.

The BMW Group's AI transformation is already having an impact across all business units, changing processes, roles, and the way the company is managed. To scale AI agents securely and efficiently, an AI-native operating model with a standardized technical layer and robust governance structures is needed. Clear guidelines for employees, consistent rights management, and continuous monitoring of token consumption create transparency, security, and centralized control, forming the basis for sustainable value creation.

By 2027, agentic AI will be an integral part of business processes. However, successful integration will only occur if AI agents scale securely, economically, and reliably. In their joint keynote address, Martin Stolberg, Head of AI at Sopra Steria, and Prof. Dr. Dirk Hecker, Director of the Fraunhofer Institute for Intelligent Analysis and Information Systems IAIS, will demonstrate the strategic decisions companies must make today to achieve this.

The focus is on three key areas: Speed through governance, meaning standards and clear guidelines that enable rapid implementation instead of hindering innovation; an economically viable AI setup that considers expected operating and token costs as well as an intelligent architecture and the selection of the most suitable models; and finally, digital sovereignty through the targeted use of European open-weight models to limit technological dependencies.

With SOOFI, the industrial AI model developed by Fraunhofer IAIS, the presentation focuses on a specific European technology. SOOFI demonstrates the potential of sovereign AI models that deploy agentic systems in business-critical processes while maintaining control over models, data, costs, and regulatory requirements.

AI agents should act – but not unchecked. How do companies define governance models that enable autonomy while simultaneously limiting risks? In this panel, experienced executives discuss key questions: Where are approvals and control points needed – and where does control become a hindrance? How is accountability ensured when systems act independently? And who bears the responsibility when an AI agent makes a decision? Learn what effective guardrails look like.

Stephan Scheuer in the discussion with

Further speakers are being coordinated.

• What AI skills will managers, departments, and technology teams need in the future?
• How can further education be transformed from a voluntary offering into an integral part of daily work?
• How can companies prevent different approaches to AI from exacerbating existing inequalities and resistance to transformation?

Lina Knees in the discussion with

• Which skills and capabilities do employees need today — and which will matter tomorrow?
• How can HR strategies position employees as active shapers of the transformation — not just recipients of change?
• How can companies build AI-ready teams and enable employees to use AI confidently, responsibly and productively?

Larissa Holzki in an interview with

What remains of a conference day? Andreas Gärtner captured the conversations, presentations, and insights throughout the day in live images. To conclude, he condenses the key ideas into a brief recap and showcases visual highlights from the first day. A finale that brings the day's themes to life once more in a fresh way.

• What actually makes a database „AI-ready“ – beyond cleansing and central storage?
• How can companies overcome fragmented data landscapes without waiting for years of transformation programs?
• What responsibility do departments, IT, and management each bear for data quality and usability?

Lina Knees in the discussion with

Every company is currently experimenting with copilots and language models – but despite enormous investments, a real ROI often fails to materialize. The reason: Consumer AI knows everything about world history, but nothing about how your business actually operates. Without deep operational context, AI models search for answers in countless IT silos – leading to inaccurate results and causing token costs to skyrocket.

FRITZ! has stood for digital sovereignty and "Made in Europe" for decades. The rapid development of AI now presents Europe with structural questions: How can AI be implemented within the tension between international competitive pressure, dependencies, and regulation? What framework conditions does Europe need to avoid falling behind in AI value creation – and which mistakes of the Big Data era must be avoided this time? Jan Oetjen shares practical perspectives and discusses what companies and policymakers must do now so that European firms can proactively shape, rather than reactively influence, the AI landscape.

• Which platform architecture enables rapid experimentation without creating new shadow IT?
• How much standardization is needed for scaling? Where must there be room for decentralized innovation?
• How can companies avoid new dependencies on individual cloud, model, or platform providers?

Lina Knees in the discussion with

How is AI changing cyber defense in companies? Stefan Pütz from Deutsche Telekom and Jasper von Hoersten from OpenAI discuss, based on concrete practical experience, how AI models are used for cyber defense to identify vulnerabilities and develop patches. The focus is on the insights Deutsche Telekom has already gained and what companies can learn from this for their security strategy and the operational use of AI.

Larissa Holzki in conversation with

When AI operates in robots and autonomous systems, digital decisions impact the real world. Where do companies see meaningful applications – and what still needs to be addressed for reliable operation? In this meet-up, participants will share experiences and questions regarding security, integration, economic viability, and the transition from testing to everyday use.

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

Not a scare tactic, but a practical assessment: What incidents involving AI agents actually occurred this year, why the fear surrounding these large-scale models misses the point when equivalent capabilities are readily available, and why the gap in most companies isn't a lack of knowledge, but a lack of implementation. With concrete answers to where companies need to focus their efforts now – both internally and externally.

Autonomous AI agents: Between the promise of productivity and loss of control
Companies are deploying AI agents that independently conduct research, operate systems, and prepare decisions. At the same time, the incidents of recent months demonstrate that this same autonomy is pushing boundaries: agents from test environments of leading AI labs have unintentionally attacked real-world corporate systems – without authorization, without anyone initially noticing, and without it being clear who is liable. The discussion about regulation is coming too late, when powerful models are already freely available.

We don't want to discuss whether AI is dangerous, but rather what decision-makers can do now: How does a company maintain control over AI that operates internally with extensive rights and is also used externally as an attack tool against the same systems? What responsibility do providers bear, and what responsibility do the companies themselves bear? And what changes for management and supervisory duties when vulnerabilities are found in hours instead of months?

• Who is liable if an AI agent causes damage without a human order – the provider, the operator, or no one?
• Is traditional IT security still sufficient when the new insider is not a person, but a co-pilot with access to everything?
• What does it mean for due diligence if every gap can be found by anyone within hours?
• Regulate or shape: What can companies retain control over themselves?

Larissa Holzki in the discussion with

Those who don't control their agents will be controlled by them: 278 keys, 42 systems, 5 days – a practical report

Your AI agents likely have more rights than most of your executives. Why do so many companies miss their AI cost forecasts? Who decides what an agent is allowed to do independently? These questions are central to the Summit. The Masterclass answers them where control originates technically: in the layer between the agents and the systems that use them.

Postman operates around 70 AI agents across 42 production systems and had to manage three things to achieve this. First, access: keys remain in the vault, permissions are per resource rather than per provider, and each request is assigned to a human, an agent, and a run. Second, costs: token budgets for humans and agents, with full transparency about who consumes what and for what purpose. Third, connectivity: what has proven effective when agents access enterprise APIs via MCP, and where MCP servers are becoming the new shadow IT.

You will take with you: a three-part control checklist (access, costs, connectivity) that you can use on Monday to check where your agents are currently operating uncontrolled.

• Who in your company can say today which agent did what yesterday, and on whose behalf?
• How do you manage token budgets for people and agents, and how do you make it visible who consumes what and for what purpose?
• How can each call be assigned to a person, an agent, and a run?
• One million tokens or one thousand: Which MCP architecture proves its worth, and where does it become the new shadow IT?.
• How many master keys to your production systems are currently stored on laptops in your company? And how quickly does one of them end up on the internet?

Your AI agents are running. But nobody sees how well they're performing, and the token bill ends up with you. Many companies use the largest language model for every task because nobody can measure whether a smaller one would suffice.

This session presents a real-world case: A single agent application was incurring over €350,000 in token costs annually. Ongoing quality assurance revealed the agent's shortcomings and determined which model was best suited for each task. The result: 42% less token cost while maintaining quality.

We'll show you live how an agent is vetted step by step. And we'll answer the underlying question: Who in your company is actually responsible for agent quality?

They will take with them:

• How to make token costs per agent visible and lower
• How real-time agent quality assurance works
• When a smaller model is sufficient and when it is not
• What new role your organization needs for this

AI is leaving the screen and becoming physical. When software meets robotics, new production logics, new value chains, and new demands on labor and skilled workers emerge. Who decides on the data and models in this process? And what does it mean for Europe when key parts of this value creation originate outside the continent?

Lena-Sophie Müller, CEO of the D21 e. V. initiative, shows which paradigm shifts Physical AI is triggering, why data is becoming a decisive competitive factor and what framework conditions politics and companies now need to create.

How is AI changing security issues in companies – and how can innovation be enabled without exposing data and systems to unnecessary risks? Exchange ideas with other participants about access rights, secure usage, protective measures, and open challenges. The focus is on practical experience and where organizations can specifically implement changes.

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

On November 17th, Nuwacom will open its doors to selected guests for an exclusive Executive Business Lunch. In a relaxed atmosphere during the lunch break, you will have the opportunity to network with executives and decision-makers from various industries.

Interested in participating? Please direct your inquiries to A.Baffour@extern.handelsblattgroup.com.

Sascha Böhr (Co-Founder, Nuwacom) is there as host and is looking forward to the exchange at the table.

Robotics and autonomous systems are becoming the means of execution in the real world – meaning the legal framework determines scalability, time-to-market, and investment security. Dr. Alexander Baghal-Schmid's presentation clarifies which regulations apply simultaneously to Physical AI and which deadlines for 2026–2028 now need to be included in roadmaps and product decisions. Specifically: What prerequisites must companies and manufacturers create now to ensure that Physical AI moves from pilot projects to secure, compliant, and regular operation?

On the new Start-up Stage During breaks at the Handelsblatt AI Summit, young teams will showcase the possibilities of AI for tomorrow. On both days, four selected startups will pitch their ideas in three- to five-minute presentations to decision-makers from business, technology, and innovation, followed by a Q&A session with Moritz Schönleber.

How do women experience the development and use of AI – and what is needed to ensure that more perspectives are involved in shaping this transformation? This peer-to-peer exchange focuses on experiences, visibility, career paths, and influence in the AI industry. Get to know other women working in AI, share perspectives, and build new connections.

All meet-ups are designed as 30-minute peer-to-peer exchanges: no presentations, but rather an unmoderated exchange among participants. Share experiences, compare different approaches, and take away inspiration for your own organization.

Participation in this event is limited to 25 people. Please register in advance via the program app so we can reserve a place for you.

How can a company with 600 employees compete against tech giants that are ten to twenty times larger? Instead of scaling the number of employees, it needs a fundamental rethink of how talent uses AI to achieve exponential impact.

In this keynote, trivago CEO Johannes Thomas goes beyond the AI hype and asks the more crucial question: What does it really take for genuine organizational transformation? trivago's shift from AI-supported work to workflows fundamentally restructured by AI forms the framework of a strategic case study – one that goes beyond simply introducing AI tools and rethinks how an organization works and what it takes for every individual to operate at a fundamentally higher level.

Efficiency is only one part of the added value that an AI transformation enables. The crucial lever behind 10x impact lies in enabling each individual to become significantly better at the work that truly matters: building systems, managing quality at scale, deepening the relationships that drive value creation, and making better decisions faster and with greater confidence.

From autonomous workflow to the agentic economy

AI agents are evolving from assistant systems to independent actors. They analyze information, make decisions, initiate processes, and could interact and negotiate with customers, suppliers, or other agents in the future. They are increasingly taking over tasks that were previously reserved for human employees.

This raises not only the question of what is already technologically possible. Companies must clarify which decisions they can delegate to agents, how multiple agents can collaborate, and who bears the responsibility if an autonomous system makes mistakes or triggers far-reaching consequences.

It is also about how companies must deal with this new form of autonomy, who bears responsibility, and whether agents will change markets and business models as drastically in the future as platforms and digital ecosystems do today.

• What can AI agents actually decide and execute independently today?
• When does genuine business value arise, and when does autonomy become a risk?
• Why is deployment in established companies particularly challenging?
• What changes when agents interact and negotiate with customers, suppliers, or other agents?
• How must companies change their processes, business models, and responsibility structures to avoid becoming mere suppliers for new agent ecosystems?
• Who bears the responsibility when an agent makes a fateful decision?

The Handelsblatt editorial team in discussion with:

Further speakers are being coordinated.

While companies invest billions in artificial intelligence, another form of intelligence is being overlooked: human intelligence (MI).

AI can analyze, formulate, and prepare decisions. But it encounters organizations where human judgment, imagination, and experience are often given surprisingly little space. The paradoxical danger: We are investing in increasingly intelligent machines – and simultaneously removing human intelligence from the workplace.

How can organizations bring AI and MI together in a new way? The future belongs neither to humans nor to machines alone. It belongs to organizations that understand when machines calculate better, when humans judge better, and how artificial and human intelligence combine to create real performance.

The crucial question is therefore no longer just: How intelligent will AI be? But rather: How revolutionary would it be to unleash AI on the world of work now?

How companies can manage the uncontrolled use of AI without stifling innovation.

Shadow AI isn't just a security issue. It's a symptom. It reveals that the official organization isn't AI-ready. That processes are too slow. That employees think faster than their companies can act. Those who merely suppress shadow AI also suppress the innovation that employees are already driving. Those who ignore it ignore the risks that are growing daily. In this panel, we'll talk to leaders managing this complex situation – about the reality behind the policies, the hidden risks, and the question: How do you build a culture where AI innovation is possible – but controlled?

Luisa Bomke in the discussion with

• Physical AI connects machine learning models with machines that operate in the real world. The crucial question is not whether impressive demos are possible, but when systems can be used reliably, safely, and economically in production and logistics.
• What advances in robotics basic models, sensor technology, and learning methods are changing industrial applications?
• Where does the greatest economic leverage lie: flexibility, productivity, skills shortage, or new products?
• What is missing between successful demonstration and scalable operation – data, standards, security, hardware, or business models?

Lina Knees in the discussion with

AI is changing how leaders work, think and make decisions. But how can individual AI use create value beyond personal productivity?

In this panel, business leaders will discuss how they are turning their own experiences with AI into broader organizational change. As AI agents begin to take on tasks, coordinate processes and act with greater autonomy, companies must rethink how work is organized, how decisions are made and where responsibility lies. The panel will examine what it takes to move from individual experimentation to secure, scalable and measurable business impact – including the governance, skills and operating models required for agentic AI.

The panel will offer practical insights into how leaders can move from using AI themselves to enabling their organizations to work, decide and create value in new ways.

Stephan Scheuer in the discussion with

How would we organize work if we were starting over today? AI can do far more than just write texts or summarize information. With agent-based workflows, it's beginning to actually take over tasks. This changes the rules of the game for companies and for us as individuals. But many are missing out on a large part of this potential if they simply apply AI to existing processes instead of rethinking them from the ground up.

Christoph Magnussen shows how companies can now make the leap from generative to agentic AI and why we need a new understanding of work to do so. Human expertise and craftsmanship will not become less important, but rather a prerequisite for using AI effectively.

What remains of the second day of the conference? Andreas Gärtner captured the conversations, presentations, and insights throughout the day in live images. To conclude, he condenses the key ideas into a brief recap and showcases visual highlights of the second day. A finale that brings the day's themes to life once more in a different way.