Lenovo brings Tianxi AI 4.5 to AI PCs with an execution-first 'Super Mode'

Lenovo released Tianxi AI 4.5 for its AI PCs on October 30. According to Lenovo-supplied material republished by QbitAI, the same version had already gone live on its AI phones on October 20 and its AI tablets on October 27, so the PC launch completes coverage across its three main device lines.

The centerpiece is a feature Lenovo calls Super Mode, which moves the assistant from answering questions to carrying out work. The company says it has connected to top applications including Feishu, Lenovo's browser and WeCom, and merged those connectors with a personal knowledge base, so a single spoken or typed instruction can be broken into steps and executed across apps: checking meetings, assigning tasks, organizing to-dos, or searching several platforms and sending out notices.

Alongside it sits a skills marketplace that gathers connectors, multi-task coordination and assorted skills, with presentation generation among the capabilities Lenovo says it improved. A separate system-control package goes further: a user can ask the assistant to put the machine in its best state and it will apply a combination of graphics, power and system optimizations. That kind of system-level access is precisely what separates an AI PC from a purely cloud-based assistant.

For meetings, Lenovo rebuilt its note-taking and knowledge tools. During a call the assistant produces live transcription and simultaneous-interpretation-grade captions, and can join as a participant that offers a third-party viewpoint or adds ideas; attendees can also question it mid-session and get answers grounded in the live transcript, with key points flagged. Afterwards it generates a summary card, files it and syncs it across devices, while the knowledge base now accepts up to 20 uploaded files per query and allows images to be dropped into the chat for mixed document-and-visual analysis.

Travel and commute work is handled through interaction design. All three device types support full-duplex voice, so there is no repeated wake word and a user can interrupt or add instructions mid-sentence. When hands are busy, the AI phone and tablet act as a remote console that can trigger Super Mode on an office or home AI PC to continue a heavy task, with progress and completion pushed as notifications and outputs previewable and manageable on the mobile device.

For heavier workloads, Lenovo's AI host machines also move to Tianxi AI 4.5, which the company says can work through complex tasks around the clock. The AI host P7 can call cloud models or function as an edge compute device on its own, and a new Auto mode matches tasks to cloud or local models based on complexity to cut unnecessary consumption.

Taken together, this is Lenovo's one-core, many-devices pitch: Tianxi AI as the core, connecting AI PCs, tablets, phones, AIoT devices and AI hosts across a device-edge-cloud architecture. What actually differentiates an AI PC is not a benchmark score but whether the assistant can be granted system permissions and finish cross-application work on the user's behalf.

That is also where the questions sit. The connectors reach into third-party accounts and data, and system control touches graphics and power settings, so permission boundaries, data flows and the ability to roll back a mistaken action will determine whether users leave Super Mode switched on. Watch how fast the connector and skills catalogues grow, and whether Lenovo publishes visible constraints such as permission tiers and action logs.