In the past two months, Taobao has accelerated the AI transformation of its search ecosystem.
In August, Taobao quietly rolled out its “AI Universal Search” across the platform. By September, internal tests for “AI Assistant” were underway, and “SnapSearch AI” for finding lower prices was fully deployed. These three AI-powered tools cover traditional search, image search, and post-search interactions, forming a comprehensive search capability matrix on the platform.
We recently gained internal access to the AI Assistant, giving us a unique opportunity to analyze how Taobao is leveraging AI for search and how it can deliver practical results for users.
AI Universal Search: Ideal for Discovery and Inspiration
“AI Universal Search” is perhaps the most intuitive application of large language models on Taobao.
Placed modestly on the secondary search page, it does not interfere with users who know exactly what they want—they can continue using traditional search seamlessly. But for users with vague or exploratory intent, the Universal Search becomes a powerful alternative. For example, searching for “Mid-Autumn Festival gifts” is a classic case of ambiguous intent.
In essence, AI Universal Search extends the user’s search experience. Clear intent leads to traditional search, while exploratory queries are funneled into AI Universal Search, creating a natural pathway into AI-driven recommendations.
During testing, AI Universal Search excelled in early-stage filtering, helping users narrow down options. For instance, entering “What to gift my girlfriend’s parents for Mid-Autumn Festival?” triggers the AI to consider multiple dimensions like occasion, recipient, and budget. It then presents three major categories: “Health & Wellness,” “Tea & Tea Sets,” and “Massage Devices.”
Users can click on a category to jump directly to detailed product listings. Multi-turn interactions allow further refinement—for example:
“My parents have diabetes and don’t like tea.”
The system will automatically remove tea-related suggestions, highlighting suitable health-focused gifts like sugar-free mooncakes. Users can continue to filter by price, origin, and other preferences.
Compared to traditional search, which simply returns products based on historical data, AI Universal Search offers nuanced, actionable guidance. It also excels in product comparisons. Asking, “iPhone 16 Pro vs. iPhone 17, which to choose?”, generates a detailed comparison covering key specs, recommended user groups, and price ranges—far more informative than the basic in-platform comparison tools.
AI Assistant: Your Personalized E-Commerce Guide
While AI Universal Search is excellent for research, AI Assistant functions as an intelligent shopping guide within traditional search.
For example, if you’ve decided to buy an iPhone 17 but are concerned about scratches, you might search for a protective case. The system presents basic categories by style and brand. By clicking “Let AI Choose”, the AI Assistant generates a refined selection interface:
-
Five major categories
-
Over 19 options
-
Five follow-up prompts
Users can select desired features, and the system dynamically filters products. A smooth animation transitions non-relevant items out, leaving only the best matches, creating a satisfying “sifting” experience.
The AI supports multi-turn dialogue, allowing detailed requests like:
“No silicone, must not yellow, with stand.”
This transforms the tool from a simple filter into a conversational guide. The more non-standard or niche the product, the more powerful the AI Assistant becomes. For instance, searching for stress-relief toys triggers the AI to consider multiple dimensions—material, use case, and functionality—helping users make decisions they might not have been able to on their own.
In categories like sports equipment, where brands and models can overwhelm consumers, AI Assistant acts as a one-click personal guide, bypassing official recommendations while leveraging real-time price updates on the platform.
Taobao also integrates AI Assistant cards directly into product feeds, proactively suggesting relevant items if the system detects that users are struggling to find suitable products.
SnapSearch AI: Bringing Physical Products Online
While AI Universal Search helps users “take the first step” and AI Assistant helps them “refine choices”, SnapSearch AI converts physical product interactions into online discovery.
Users can snap a photo of a product in the real world, and SnapSearch AI identifies it, offering:
-
Lowest price options
-
High-value alternatives
-
Related products
For example, photographing a pet dental chew or a shampoo bottle generates not only the exact match but also suggestions for alternatives and different sizes. AI’s generative thinking ensures it constantly expands on potential user needs, turning a traditional search tool into an inspiration-driven entry point.
Key Takeaways for Global E-Commerce
Taobao’s AI initiatives demonstrate scenario-driven AI deployment rather than an end-to-end one-size-fits-all approach. By breaking down search into stages—exploratory, detailed selection, and real-world item identification—Taobao ensures AI:
-
Understands the context of each user action
-
Provides actionable guidance without interrupting the experience
-
Gradually teaches users to leverage AI
The result is a system where AI complements human behavior, not overrides it. For global e-commerce platforms, this approach offers a clear lesson: AI should be deployed to enhance real user scenarios, not just as a standalone feature.
