Tencent Releases TAI Agent Solution: Yuanbao Goes Onboard, Cockpit AI Shifts from Chatting to Getting Things Done

Over the past year, the call for AI agents to be integrated into cars has been loud in the automotive industry. Voice interaction has basically become standard on new cars, and recognition speed and accuracy have also improved. But when it comes to actually having the cockpit AI assistant handle certain matters, such as planning an itinerary, users still instinctively pick up their phones to do it.

During the 2026 Tencent Global Digital Ecosystem Summit, in conversations with Zhong Xuedan, Vice President of Tencent Cloud and Head of Smart Mobility, Li Bo, Vice President of Tencent Cloud, and Sun Yan, Vice President of Tencent Cloud, Che Dongxi found Tencent's more concrete solution to this problem.

The configuration race for smart cockpits has reached an awkward point.

Data from Gasgoo Automotive Research Institute shows that by August 2026, the penetration rate of smart cockpits in domestic passenger vehicles had reached 87.5%. Basic configurations such as central control screens and voice interaction are approaching saturation in new cars, and the configuration list itself can hardly distinguish one from another anymore—getting smart cockpits "into the car" is no longer the problem; the problem lies at the level of "actually using them."

Users' needs are actually very clear. Industry surveys show that 84% of respondents expect a proactive in-car partner, and 64.1% of new energy vehicle owners list services as a main reason for purchasing a car.

But the reality is that AI agents are crowding into cars, and most products remain at the level of "being able to chat." They can announce the weather and play music, but once it comes to tasks that require cross-system completion, such as planning an itinerary or paying parking fees, the product experience often breaks down.

One aspect is that the experience cannot form a closed loop; another major problem is insufficient reliability.

A 2026 J.D. Power study shows that among users' quality complaints about advanced vehicle configurations, "unstable and inaccurate operation" accounts for 48%; a 2025 study also found that 56.5% of users actively reduce their use of AI vehicle control due to reliability and safety concerns.

Functions are not stable enough, which affects users' willingness to use them; user usage rates are low, and a closed-loop experience is then out of the question.

Zhong Xiangping, Vice President of Tencent Group and President of Tencent Smart Mobility, believes that putting an Agent in the car is not just a chat entry point, but should become a reliable good partner and a capable good helper.

▲Zhong Xiangping, Vice President of Tencent Group and President of Tencent Smart Mobility

So how has Tencent done it?

At this conference, Tencent brought a newly upgraded TAI agent solution, composed of four parts: the TAI cockpit assistant, the TAI mobility service agent, the TAI content and entertainment agent, and the TAI cross-device continuity service.

▲Tencent releases the smart cockpit TAI agent

The TAI cockpit assistant integrates Yuanbao's full-duplex end-to-end voice, Yuanbao search, and the Hunyuan large model. Low-latency voice adapts to multiple sound zones in the cabin, and the source quality and real-time nature of Yuanbao search are used to improve the reliability of answers; cross-device memory and context awareness allow the cockpit assistant to understand scenarios and user intent and to coordinate mobility, entertainment, and vehicle control tasks across domains.

Simply put, after Yuanbao is integrated, the cockpit assistant can base every answer on evidence, making answers more reliable.

After answering accurately, users also hope the cockpit agent can get things done. The TAI mobility service agent connects Tencent Maps, WeChat Mini Programs, and WeChat Pay, supporting voice booking of hotels, buying tickets, adjusting itineraries, providing guided commentary along the way, and AI outbound calls to restaurants for reservations.

Content-level needs are handled by the TAI content and entertainment agent, which includes music, video, news, podcasts, image generation, image recognition, and other services. The AI agent is not only the user's "chat buddy," but can also continuously provide companionship and services to users during vehicle use.

Moreover, Tencent has extended the cockpit agent's services from inside the car to outside the car. The TAI cross-device continuity service connects to consumer entry points such as the Yuanbao App, Xiaowei, and Tencent Maps. Users can check vehicle status, control the vehicle, transfer addresses to the car, and synchronize itinerary information in WeChat, achieving "phone-car" service connectivity. Services across all scenarios before, during, and after the trip remain uninterrupted, with the same assistant and the same set of memory, realizing service continuity.

Li Bo, Vice President of Tencent Cloud, summarized this combination as: "An Agent assistant that can chat with you, get things done, and stay by the user's side both inside and outside the car."

▲Li Bo, Vice President of Tencent Cloud

Overall, putting Yuanbao in the car is pushing the competitive focus of smart cockpits to a new height: from "whether there is an Agent" to "whether the Agent can get things done within safety boundaries."

At present, Tencent's smart cockpit solution has been mass-produced and deployed in automakers such as Changan, NIO, and Leapmotor, with more than 20 million vehicles equipped.

Beyond evolution on the product side, Tencent has placed another main line on the enterprise side.

At present, integration in China's automotive industry is a major trend, and many automakers have initiated organizational structure adjustments, with one of the core purposes being organizational efficiency improvement.

Where exactly is this process stuck? People from the front line are most qualified to speak.

At the meeting, an industry executive revealed that in the process of integrating AI into business, one of the difficulties is how to clearly explain what the business is actually supposed to do, and how to truly use a demo in workshops, bases, and production lines. How to fill in the missing context and the tacit knowledge in experts' minds?

English text is being prepared.

Tencent has also built a series of efficiency tools: WorkBuddy (an AI office agent) and CodeBuddy (an AI coding assistant) are aimed at improving employee efficiency, while the enterprise Agent platform targets long-running, multi-person collaborative business processes. Working together, the two distill methods proven effective by individuals into skills and expert capabilities that can be reused across the organization.

Beyond tools, Tencent uses an FDE (Forward Deployed Engineer) co-creation model with businesses, bringing together technical staff, business consultants, and automakers' business teams to jointly select high-value scenarios, define the boundaries of human-machine collaboration, risk control, and evaluation metrics, and embed agents into business processes rather than leaving them suspended at the tool layer.

The results of this approach are already evident. The "AI Rocket Class" jointly built by Tencent and Changan Automobile has launched and is operating eight major scenario agents covering product planning, production-sales coordination, quality operations, AI Coding, and other areas, with efficiency greatly improved.

Tencent and a leading automaker have developed dozens of agents on Tencent ADP. A vehicle model planning assistant reduced the time for preliminary data analysis from weeks to hours, and improved the efficiency of research and planning reports by over 95%. Tencent also worked with a new luxury EV brand on Tencent CNB and TokenHub, reducing software bugs by over 90%.

Tencent's series of solutions has always centered on the enterprise's overall position: no longer just improving efficiency for a single role, but driving organizational transformation, ultimately forming a super organization that is agile and flat, highly collaborative, and intelligently driven by human-machine collaboration.

Sun Yan, Vice President of Tencent Cloud, said that currently nearly 90% of domestic automakers use Tencent's AI agent products and services such as WorkBuddy, distilling individual practices into enterprise knowledge and skills and turning them into enterprise assets.

▲Sun Yan, Vice President of Tencent Cloud

In addition to the two main threads of cockpit and organizational efficiency, Tencent also upgraded its overseas intelligent solutions and new-scenario solutions at this event.

On overseas expansion, building on its overseas basic cloud and connected vehicle cloud solutions, Tencent upgraded its solution set covering four major scenarios: overseas intelligent driving, marketing services, overseas maps, and efficiency agents.

As an extension of intelligent assisted driving capabilities, Tencent is also continuing to strengthen exploration in two new scenarios: embodied intelligence and Robotaxi. It launched an embodied data closed-loop suite that, combined with AI infrastructure and the Tairos embodied intelligence open platform, accelerates the validation and deployment of embodied intelligence in factories, warehousing, and other scenarios.

In the Robotaxi field, Tencent has made clear that it will not do Robotaxi itself. Instead, it will use its consumer service entry point and cloud-map capabilities to connect intelligent driving R&D, fleet operations, and user mobility, and use cloud-map capabilities and mobility ecosystem services to support fleet R&D and large-scale operations. In Li Bo's words, what Tencent wants to be is "the cloud behind every Robotaxi fleet."

Conclusion: Stay out of the foreground and be a good digital assistant

From putting Yuanbao in cars and building the smart cockpit TAI agent, to launching a full-scenario enterprise agent solution, to extending capabilities into new overseas markets and new scenarios such as embodied intelligence and Robotaxi, Tencent's layout in the automotive industry looks broad, but its main thread is very clear: based on the same cloud + AI foundation, it sticks to its positioning as a digital assistant that does not build cars and does not enter the arena.

Automakers are responsible for building good cars and creating brand and product differentiation, while Tencent provides the AI infra platform capabilities that automakers do not want to invest in repeatedly, rather than competing with them for the right to define.

At the event, Zhong Xiangping gave a clear explanation of Tencent's industry role in the AI era: "We insist on being a good digital assistant. Automakers are responsible for building good cars, and we are responsible for making AI usable and for achieving better results with it." Staying out of the foreground, not seizing the right to define, and folding AI capabilities into automakers' products, processes, and scenarios so that technology ultimately lands where users can perceive it—as the intelligent vehicle industry develops in depth, the value of Tencent as a digital assistant is being reassessed.