Carsten Tusk, co-founder at Xyonix, shares his perspective on the evolving AI landscape, highlighting the rapid progress in multimodal models and on-device intelligence. He foresees a blend of general-purpose systems with industry-specific solutions for deeper impact. Companies are now focusing on advanced use cases like custom AI assistants, digital twins, and real-time edge applications. He emphasizes that success in AI depends on strong data foundations, clear objectives, and cross-functional collaboration. Looking ahead, he sees AI as a collaborative co-pilot, seamlessly supporting human creativity and decision-making across key industries.
What trends in AI are you most excited about right now?
I’m particularly excited by the rapid advances in multimodal foundation models that seamlessly integrate vision, language, and even audio. Developments in self-supervised learning are unlocking richer feature representations without massive labeled datasets. I’m also intrigued by the emergence of lightweight, on-device models that bring powerful AI capabilities directly to end users. Finally, the growing ecosystem of AI-native tools and agent frameworks is making it easier to build sophisticated, autonomous workflows.
Do you think AI will become more specialized by industry, or continue to be used across everything?
I see a hybrid landscape emerging: broad, versatile backbone models will provide general-purpose capabilities, while fine-tuned, domain-specific variants will address industry-specific challenges. Vertical specialization—say in legal, medical imaging, or financial forecasting—will drive deeper performance and compliance. At the same time, cross-industry interoperability and API-driven services will ensure that foundational AI remains broadly applicable. This balance allows organizations to leverage both general models and tailored solutions.
What kinds of AI projects are companies asking for more now compared to a year or two ago?
Whereas earlier requests centered on basic chatbots and predictive analytics, I’m now seeing a shift toward retrieval-augmented generation, custom knowledge-base assistants, and AI-driven process automation. There’s growing interest in embedding models into edge devices for real-time inference, as well as in governance and monitoring platforms to ensure model robustness in production. Companies are also exploring AI-powered simulation and digital twin projects to optimize complex systems. In gaming, there’s renewed focus on procedural content generation and intelligent NPC behaviors using LLMs.
How do you see people and AI working together in the future?
I envision a collaborative “co-pilot” paradigm where AI handles routine, data-intensive tasks and suggests creative directions, while humans focus on oversight, ethical judgment, and strategic decision-making. AI will act as an adaptive assistant that learns each user’s preferences and workflows, enabling tighter human-in-the-loop feedback cycles. In creative fields like game development, designers will sketch ideas in natural language and let AI prototype levels or character designs. Ultimately, the most productive teams will integrate AI seamlessly into daily tools, not view it as a separate silo.
Which industries do you think will benefit most from AI in the next 2–3 years?
Healthcare stands to gain enormously through AI-assisted diagnostics, personalized treatment planning, and de-risked drug discovery. Manufacturing and logistics will see productivity boosts from predictive maintenance, supply-chain optimization, and AI-driven robotics. Financial services will leverage AI for fraud detection, risk management, and personalized advisory services.
