Staff Reporter |
News Editor | Wen Shuqi
On August 20th, Baidu’s newly-appointed head of the APP R&D Center, He Jingzhou, made his first public appearance, unveiling two major upgrades to the company’s search product line.
One of these upgrades is a new search results format called “GUI General Knowledge Cards.” During a hands-on demo, a Baidu product manager told us this represents their pioneering “interactive full-page” search format. It cleverly merges knowledge cards, step-by-step diagrams, and interactive animations into a single, unified page. Right now, it covers all the high-frequency exam points in physics, chemistry, and math.
We tried it out ourselves. Just type “Pythagorean theorem” into the Baidu APP, and the “GUI General Knowledge Card” pops up. It visually breaks down the theorem’s principles, letting you drag and drop the triangle’s legs to see how the math works in real time. And if you search for “Wenxin Teacher,” you can trigger a photo-solving mode. An AI teacher then guides you through the problem step-by-step, just like a human tutor, and you can even interrupt to ask questions whenever you’re stuck.
The second product is the Wenxin Assistant Task Engine 2.0. This handy tool can whip up PPTs, draft analysis reports, and handle scheduled tasks. Its key differentiator in the market? It’s completely free.
He Jingzhou explained to us and other media at the event that in the age of AI, search results shouldn’t just be a static list of links. Baidu’s vision is to hand users interactive, usable tools and keep innovating from there.
This push for innovation is a direct response to the massive disruption caused by AI assistants like Doubao and Qianwen. The entire framework of search—from user interaction and traffic distribution to the underlying business model—is being structurally rebuilt. The latest earnings report shows the pressure is real: Baidu APP’s monthly active users (MAUs) hit 644 million in Q2 of this year. That’s a drop of 91 million year-over-year and 11 million from the first quarter.
The user exodus is just one headache. The more alarming issue for Baidu is the decline in search ad revenue. The same financial report revealed that its online marketing revenue fell to 13.1 billion yuan in Q2, a steep 19% drop year-over-year.
But are these product tweaks enough to turn the tide? A search industry insider we spoke with is skeptical, suggesting these upgrades address peripheral needs, not the core reasons people use search. It feels a bit like “change for the sake of change,” he said, as if Baidu is just chasing whatever trend is hot at the moment.
Baidu has been aggressively experimenting with search overhauls since last year. The core strategy, according to a person familiar with the company, is to fuse large language models with search to enhance user experience.
At the time, Zhao Shiqi, then GM of Baidu Search, championed supporting longer, vaguer queries. The idea was to go beyond simply listing existing info and to actually help users get things done—checking the weather, booking hotels—and to use AIGC to fill the content ecosystem gaps that had plagued Baidu during the mobile internet era.
Zhao led the most significant revamp of Baidu’s search product in a decade. But before his ambitious plans for content and services could fully materialize, he left the company in March of this year.
His successor, He Jingzhou, previously focused on large model algorithms at Baidu, signaling that the company’s core strategy remains the fusion of LLMs and search. Judging by the recent communications, though, He’s approach is less about inventing new user habits and more about simplifying the path for existing demands.
The product insider we spoke with argues that Baidu’s problems aren’t solely AI-driven. The deteriorating quality of its own search results and users flocking to platforms like Xiaohongshu for content discovery all play a part. “It’s a multi-factor problem that can’t be reversed by just tweaking a product feature,” he said.
Adding to the challenge, the search engine itself is a mature product with limited room for radical change.
Globally, even Google, which has been embedding AI into its search, hasn’t fundamentally altered its core product or business model. The search box, keyword matching, results pages, and the ad system still form the backbone of its offering.
Back in China, the only major players still heavily investing in search engines are Baidu and Alibaba’s Quark. When faced with the AI onslaught, Quark chose a more restrained path. Instead of a radical overhaul, they simply let users toggle seamlessly between the traditional search box and AI assistant features.
An insider at Quark told us that while the impact of AI assistants is undeniable, their internal assessment is that search won’t be entirely replaced. “For straightforward fact-finding, an AI assistant is quicker,” he explained. “But for complex tasks—like transferring money, making a payment, or booking a trip—users still prefer to search their way to the right website and handle it themselves. A search engine should just excel at fulfilling that need.”
But Baidu’s constant restructuring suggests they’re not satisfied with just being a utility. They still harbor ambitions of becoming the ultimate, all-encompassing traffic gateway for the AI era—the starting point for every answer, website, service, or booking.
That grand vision, however, is increasingly difficult to achieve through product iteration alone in today’s market. Baidu seems to be betting on organizational fixes, hoping a leadership change or a new business structure will somehow unlock a breakthrough.
This ongoing “tinkering” isn’t about accelerating down a clear path. It’s a sign that Baidu doesn’t actually know how to build a comprehensive search entry point for the AI age. And that, more than anything, is the biggest dilemma facing Baidu Search right now.
Until Baidu figures out that answer, it will continue to suffer the double whammy of user attrition and falling ad revenue.
