The case for Hybrid AI: personalized, active intelligence to understand you and the moment

Lenovo argues that the future of AI is hybrid, combining cloud intelligence with on-device processing to handle sensitive data, reduce latency, and lower costs. The company says cloud-only AI is impractical for personalized, proactive tasks that depend on local context and real-time responses.

Most AI solutions start with a prompt. Some days move at an unforgiving pace: a customer meeting, an investor briefing, a partner discussion, and a media conversation. The names, priorities, follow-ups, sensitivities, and open questions can start to blur. But the expectations do not. Each person expects me to remember the conversation, understand their context, and be ready for what matters to them. That is exactly the kind of moment when AI should be most useful. Before I walk into the room or jump on the call, I don’t need a generic answer. I need a briefing that understands the situation. What did we discuss with this customer last time? What did they ask about? Which deck did they see? What did my team promise to send? What has changed in their business since then? Which emails actually matter? What are they likely to care about most? And what are the three points I need to land? For me, this should feel like one simple request. Behind the scenes, it’s anything but simple. Some of that information is public. Some are private. Some sit in the cloud. Some sit on my PC, phone or tablet. That’s where a cloud-first approach starts to show its limits. We’ve become used to thinking of AI as something that happens somewhere else: in a remote model, in a cloud data center, behind a chat window. That model still matters. The cloud will remain indispensable for large-scale reasoning, broad knowledge, and training increasingly capable foundational models. But here’s the practical truth: Not every workload should be sent to the cloud. Some tasks are too sensitive. Some depend on local context from the devices we use every day. Some – like live translation, real-time captions, voice interactions, or camera understanding – depend on low latency, continuous sensor input, or reliable operation even when connectivity changes. And as token usage grows, the cost of cloud-based AI can become a more visible challenge for companies and users. That’s why I believe the future of AI is hybrid https://news.lenovo.com/beyond-agentic-next-phase-of-orchestrated-ai-transformation/ . The best AI goes beyond simply answering questions and can understand the moment you’re in, and decide where intelligence should run to help you most effectively. Hybrid AI combines cloud intelligence with intelligence running directly on devices, allowing each workload to execute where it makes the most sense. Putting the right AI in the right place Hybrid AI may sound technical, but the idea is actually straightforward. It means putting the right AI workload in the right place. It’s “hybrid” because intelligence is not only in the cloud or only on your device. Yes, some requests are best handled in the cloud, where large models have access to massive specialized data centers and broad knowledge models. But others are better handled on your own devices, where AI can respond with very low latency, understand your local context, and work with information that never needs to leave your PC, smartphone, or tablet. The first generation of generative AI was largely conversational. You opened a chat window, asked a question, received an answer, and moved on. Cloud computing made sense because each interaction was relatively self-contained. The next generation of AI is different. AI is becoming more personal, more proactive, and more deeply integrated into your day to day. It’ll increasingly help manage calendars, prepare meetings, summarize conversations, search personal documents, coordinate tasks across devices, and anticipate what we need next. As AI shifts from answering questions to helping us accomplish work, a cloud-only approach becomes less practical. Many tasks depend on personal context that already exists on our own devices. Others, such as live translation, real-time captions, voice commands, or AI features inside creative apps, need to respond in the moment. Some involve sensitive business information or personal data that organizations may prefer to keep closer to the user. When AI moves closer to us As AI becomes more personal, the device becomes more important. A PC understands the work we’re doing. A smartphone offers indicators about where we are, who we’re meeting next, and what we’re doing next. A wearable may hear a question, see what’s in front of us, or capture a moment when reaching for a keyboard would feel unnatural. This is why the next generation of AI won’t just live in one form factor. PCs will remain central because so much work still happens there. Smartphones will continue to matter because they’re always with us. Tablets, wearables, smart displays, headphones, cameras, and edge devices may all become more relevant as AI starts to respond to different moments in different ways. Motorola’s Project Maxwell https://news.lenovo.com/pressroom/press-releases/hybrid-ai-personalized-perceptive-proactive-ai-portfolio-tech-world-ces-2026/ , a wearable AI perceptive companion proof of concept, offers a glimpse of where this could lead. Designed to work within Motorola Qira’s ecosystem, it imagines AI that’s always accessible and context-aware, rather than confined to a single screen or chat window. The Lenovo AI Glasses concept https://news.lenovo.com/pressroom/press-releases/lenovo-reimagines-concept-at-ces-2026/ also brings personal AI into a hands-free wearable form, while Project Kubit https://news.lenovo.com/pressroom/press-releases/lenovo-reimagines-concept-at-ces-2026/ envisions a personal AI hub that can draw from information across a user’s device ecosystem and the way they interact with their technology over time to deliver more personal AI assistance through voice or touch. As AI becomes more personal, it also becomes more physical. It needs to show up where people are working and living. The devices that already surround us may therefore play a larger role in how AI understands context, protects privacy, reduces latency, and delivers help at the right moment. Intelligence is only half the equation As AI becomes part of more workflows, another challenge starts to emerge: economics. For individual users, the costs may not always be visible. For organizations deploying AI across thousands of employees and millions of workflows, they become much harder to ignore. The issue becomes especially clear as AI moves from simple prompts to more complex workflows: agents that make multiple model calls, assistants that search across enterprise documents, coding tools that iterate through problems, customer service systems that handle long conversations, and meeting assistants that summarize, retrieve, and follow up across the workday. If every task, regardless of its complexity, is sent to the largest cloud model available, AI becomes unnecessarily expensive. But the goal can’t be just using the biggest model every time; it has to be about using the most appropriate one. Orchestration then becomes essential. The system should determine what context is needed, which intelligence to use, where the work should happen, and how to deliver the result as one continuous experience. This is the role we envision for Lenovo Qira. Rather than acting as another AI assistant, it’s a new layer of intelligence across your devices. It understands your personal context, coordinates the right AI models and agents behind the scenes, and turns disconnected experiences into one continuous flow. Instead of asking you to decide which AI to use, Lenovo Qira simply brings you the right capability when you need it. Building the Hybrid AI future I believe that hybrid AI is a systems challenge. Delivering a seamless AI experience requires intelligence across every layer: the devices we use, the infrastructure that powers large models, the software that connects them, and the orchestration that determines where each workload belongs. Lenovo already operates across the technologies that will shape this era: PCs, smartphones, tablets, wearables, edge computing, enterprise AI infrastructure, services, and cloud partnerships. And we’re helping connect these pieces into one intelligent ecosystem, where AI can move naturally between devices, the edge, and the cloud. In the end, people aren’t going to judge AI by the size of the model behind it. They’ll judge it by whether it understands the moment they’re in, protects what matters, and helps them move forward with confidence. That’s when AI stops feeling like a tool we prompt and starts feeling like smarter technology that works naturally alongside us. Luca Rossi is President of the Intelligent Devices Group at Lenovo, leading the development of a Hybrid AI portfolio spanning PCs, smartphones, tablets, commercial solutions, and other AI products including AI software/agents and AI-native devices such as wearables and ambient AI.