Whereas most consideration within the AI race is targeted on mannequin builders and cloud platforms, Lenovo sits nearer to hundreds of thousands of customers than most corporations. Because the world’s high PC maker by quantity, Lenovo ships tens of hundreds of thousands of gadgets yearly. What it decides to ship, bundle, and combine can immediately form how AI exhibits up in lots of on a regular basis lives.
That’s what made Lenovo’s announcement right now at CES notable. At a flashy occasion on Tuesday at The Sphere in Las Vegas, it launched Qira, a system-level, cross-device AI assistant designed to dwell throughout Lenovo laptops and Motorola telephones. It’s Lenovo’s most formidable AI effort so far and a uncommon have a look at how a {hardware} big with world attain is considering integrating AI extra deeply.
Jeff Snow, Lenovo’s head of AI product, instructed me how Qira got here collectively, why the corporate is intentionally avoiding a single unique AI partnership, and what he discovered from earlier experiments like Moto AI and Microsoft’s Recall debacle.
Qira emerged from a quiet however significant inner reorganization lower than a 12 months in the past, based on Snow. Lenovo pulled AI groups out of particular person {hardware} items reminiscent of PCs, tablets, and telephones and centralized them into a brand new software-focused group that works throughout the complete firm.
For an organization lengthy optimized round {hardware} SKUs and provide chains, the transfer signaled a shift towards placing AI extra entrance and heart. “We wished a built-in cross-device intelligence that works with you all through the day, learns out of your interactions, and may act in your behalf,” Snow stated. He talked about utilizing Qira’s on-device mannequin throughout his flight to CES to assist him workshop methods to speak in regards to the information in conferences based mostly on the notes and paperwork on his PC.
“We wished a built-in cross-device intelligence that … learns out of your interactions and may act in your behalf.”
Qira is just not constructed round a single flagship AI mannequin. As an alternative, it’s modular. Below the hood, it mixes native, on-device fashions with cloud-based fashions, anchored by Microsoft and OpenAI infrastructure accessed by way of Azure. Stability AI’s diffusion mannequin can also be built-in, together with tie-ins to app-specific companions like Notion and Perplexity.
“We didn’t need to hard-code ourselves to 1 mannequin,” Snow stated. “This house is shifting too quick. Totally different duties want totally different tradeoffs round efficiency, high quality, and price.”
That stance runs counter to the push from main AI labs, a lot of which might fortunately turn into the unique intelligence layer for an organization with Lenovo’s attain. Lenovo’s view is that optionality issues extra, particularly given its management over one of many largest shopper computing distribution channels on this planet.
Snow beforehand labored on Moto AI, Motorola’s assistant, which he stated noticed excessive preliminary engagement. Greater than half of Motorola customers tried it, however retention wasn’t good. He stated that an excessive amount of of the expertise felt like prompt-based chat options individuals may already get elsewhere.
“That pushed us away from competing with chatbots,” Snow stated. “Qira is about issues chatbots can’t do, like continuity, context, and appearing immediately in your system.”
Value pressures loom over all of this.
Lenovo additionally paid shut consideration to the backlash round Microsoft’s Recall function. Snow stated Qira is designed from the outset with opt-in reminiscence, persistent indicators, and clear consumer controls. Context ingestion is optionally available. Recording is seen. Nothing is silently collected.
Value pressures loom over all of this. Reminiscence costs are rising as AI demand strains provide chains, and analysts count on PC costs to observe. Qira doesn’t increase baseline system necessities for PCs, Snow stated, however it performs finest on higher-end machines with extra RAM. Lenovo is working to carry native fashions right down to smaller reminiscence footprints, like 16 gigabytes of RAM, with out watering down the expertise.
Strategically, Lenovo sees Qira as each a retention play and a hedge in opposition to {hardware} commoditization. Within the quick time period, it hopes that tighter integration between laptops and telephones will nudge prospects to remain throughout the Lenovo ecosystem. Over the long term, Snow framed Qira as a technique to differentiate Lenovo gadgets when specs alone are not sufficient.
