Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

BYD Launches Didixia AI Super-Agent, Debuting On The Tengshi N8L

BYD has released Didixia, which it calls the automotive industry's first AI super-agent, debuting on the Tengshi N8L electric vehicle and upgrading the cockpit into an AI OS on its Xuaji 2.0 architecture. The open ecosystem platform supports A2A and MCP and already connects more than ten partner agents, with several Tengshi models set to receive OTA updates.

Published

On September 21, Chinese tech outlet QbitAI carried an announcement from BYD: the automaker has released Didixia, which it calls the auto industry's first AI super-agent, debuting on the Tengshi N8L electric vehicle. Yang Dongsheng, BYD group senior vice president and president of its automotive new technology research institute, attended the launch and gave a technical walkthrough of the system.

Didixia runs on BYD's Xuaji architecture 2.0, which integrates cabin, driving and electrical systems, and upgrades the smart cockpit into an AI OS. BYD describes a device-plus-cloud split: on-device compute handles instant vehicle control responses, while the cloud handles deeper reasoning. The system also includes multimodal perception, an on-demand tool system, a reusable skill system and a memory layer that is meant to learn the driver over time.

An open ecosystem is the centrepiece of the launch. BYD built an ecosystem agent platform that supports the A2A and MCP standards, allowing third-party AI agents and developers to plug into its agent ecosystem. BYD says Didixia already connects more than ten partner agents, which it claims is the most in the industry.

The launch partner list spans travel, entertainment, life services and general chat: hotel agents with Fliggy, navigation with Amap, car-to-home with COLMO, food delivery with Taobao Shangou, audiobooks with Ximalaya, music agents with QQ Music, Kuwo, Kugou and NetEase Cloud Music, short dramas with Yuanshen Xiaoju, lifestyle services with Alipay's Abao, and video agents with iQiyi, Youku and Bilibili. Chit-chat runs on the Doubao and Qwen large models.

Capability-wise, BYD frames Didixia as able to play, to get things done and to hold a conversation. In-cabin settings and controls respond to a single spoken instruction, and when several requests arrive at once the system breaks them down and executes them step by step. The agent ships with switchable personas, and a spoken introduction completes voiceprint registration and starts a personal memory.

The Tengshi N8L electric model is the first vehicle to carry Didixia. BYD says the second-generation Tengshi D9, the Tengshi N8L fast-charge edition, the new Tengshi Z9GT and the Tengshi N9 fast-charge edition will begin receiving over-the-air updates in sequence. The information was supplied by BYD and republished by QbitAI with permission.

The reason this matters is that cockpit competition is shifting from putting apps in cars to putting agents in them. Bringing agent interoperability protocols such as A2A and MCP into the head unit gives in-car agents a chance to reuse the phone and cloud agent ecosystems instead of rebuilding a custom integration for every service, and the ten-plus launch partners provide real-world reference points for that model.

Four things to watch: the actual OTA rollout timeline and vehicle coverage, the permission and privacy boundaries attached to each connected agent, how the on-device model performs against real in-car compute limits, and whether the open platform attracts developers beyond the launch-day roster.

Why it matters

Cockpit competition is moving from adding apps to adding agents, and BYD is using A2A and MCP to pull third-party agent ecosystems into the car. The follow-up questions are the OTA rollout schedule, permission boundaries for connected agents and how the on-device model performs in practice.

BYDAgentAutomotive
Back to realtime news

Nearby Updates

All