In the summer of 2025, heralded as the "first year of AI Agents" within the industry, global technology titans and emerging startups have converged on this frontier, vying for an early lead in the next epoch of human-computer interaction.

From OpenAI's iteration of "memory function" to Google Gemini's agent matrix, and domestic advancements in edge-side models by FaceWall Intelligence and the Institute for AI Industry Research, AI Agents are revolutionizing productivity across diverse scenarios, spanning from consumer applications to enterprise services.

Amidst this landscape, Meitu recently announced its development of an AI Agent tailored for vertical scenarios. At a pivotal juncture where AI applications transition from general to vertical applications, Meitu distinguishes itself amidst fierce competition from giants like Adobe and ByteDance, leveraging its unique expertise in image aesthetics and technological prowess.

Is Meitu's confidence in its AI Agent grounded in a robust technological foundation?

Meitu's foray into AI Agents is not a spur-of-the-moment decision. Over a decade of intensive cultivation in computer vision has endowed it with a solid technological base.

Particularly noteworthy is Meitu Imaging Lab (MT Lab), the technical cornerstone of the AI open platform, which stands at the global forefront in computer vision and artificial intelligence.

Recently, seven research achievements from Meitu Imaging Lab were selected for presentation at the prestigious international academic conferences CVPR 2025 and AAAI 2025, providing continuous technical support for Meitu's AI open platform.

Moreover, a significant breakthrough in technology implementation has occurred on the edge side. In April this year, Meitu's AI open platform underwent a major upgrade, enabling efficient realization of edge-side AI image processing functions. This provides users and the industry with visual solutions based on edge computing.

Among these capabilities, the AI removal function precisely erases specified areas with realistic and natural detail completion, supporting brushing and framing for intelligent image area positioning. The AI cutout function, leveraging high-precision intelligent segmentation, covers portraits and item icons, achieving hair-level precision. The AI expansion function allows for free image enlargement, easy size switching, and adaptability to various scenario needs.

These capabilities are already supported on the latest chip platforms such as Qualcomm 8gen4, paving the way for the edge-side deployment of AI Agents.

Additionally, Meitu's technological moat is evident in its self-developed Meitu Qixiang large model. As the first large model in Fujian Province to be recorded, this model has been upgraded to version 5.0, boasting enhanced visual expressiveness and creativity. It serves as the technical cornerstone of Meitu's products, underpinning a range of innovative functions from AI portraits to video effects.

The head of technology marketing at Meitu Imaging Lab (MT Lab) stated that Meitu has launched numerous tools to serve content production, enabling natural and realistic clothing try-on effects. In February 2025, it topped the App Store charts in 12 countries and regions, adding nearly 20 million overseas users, achieving global user acquisition and traffic diversion.

Commercialization capability is another key strength. The 2024 financial report reveals that Meitu's total revenue reached 3.341 billion yuan, a year-on-year increase of 23.9%. Among this, revenue from imaging and design products amounted to 2.09 billion yuan, a year-on-year surge of 57.1%. These figures affirm Meitu's prowess in converting AI technology into commercial value.

After vertical breakthroughs, how will Meitu navigate differentiated competition?

While technology giants fiercely compete in the general AI Agent domain, Meitu has opted for a distinct path—deeply cultivating the imaging and design field. This strategic choice is rooted in a profound understanding of industry pain points.

"Behind AI agents lies a brand-new workflow," emphasized Wu Xinhong, Chairman of Meitu, at the performance exchange meeting in March 2025. "It necessitates AI to accurately comprehend user needs and deploy its capabilities to fulfill them." For Meitu, the crux of creating an AI Agent involves refining core capabilities, making them accessible for AI cloud invocation, and integrating them into a personalized workflow.

Furthermore, industry analysts point out that Meitu's approach to creating productivity tools focuses on vertical scenario pain points, encapsulating model capabilities into products, and balancing controllability and ease of use to bolster product strength. This logic will be reprised in its newly launched AI Agent product.

The current market situation validates the rationality of this strategy. In May 2025, a report by iResearch Consulting revealed a trend of vendor differentiation in the AI application market. In the realm of AI efficiency tools, Adobe, ByteDance, Kuaishou, and Meitu emerged as prominent representatives.

Among them, in the AI image and AI video product tracks, multiple AI products under Meitu, such as MeituXiuxiu, Wink, Meitu Design Studio, and WHEE, are recognized as fresh ideas driving industry innovation.

Notably, the commercialization pace of visual modality products significantly outpaces that of language-based products, which is precisely Meitu's core advantage.

Estimates suggest that China's AI industry will maintain a compound annual growth rate of 32.1% from 2025 to 2029, anticipated to surpass one trillion yuan in market size by 2029. Concurrently, the industry's competitive landscape is undergoing dramatic changes. Starting in 2025, the trend of "land grabbing" among AI application vendors has gradually emerged, with price and traffic becoming pivotal competitive strategies.

This technological competition, spearheaded by top vendors like Adobe, ByteDance, and Meitu, is steering the industry towards a critical juncture, transitioning from the "arms race of large models" to the "decisive battle of scenario implementation."

While experiencing rapid growth, Meitu's differentiated path for its AI Agent also confronts multiple challenges. Issues such as insufficient user stickiness and limited fee sustainability for some C-end AI products persist. Additionally, the advantages of giants like ByteDance in traffic and funding cannot be overlooked.

Commercial Landscape: How Will AI Agents Reshape Meitu's Future?

The impending AI Agent is not merely a manifestation of technological innovation but also a pivotal component of Meitu's future growth strategy. From a commercialization standpoint, this product will fortify Meitu's penetration in the B-end market.

In a research report released on June 16, Guotai Junan Securities accorded Meitu an "overweight" rating for the first time, explicitly stating that as a leader in image digitization, Meitu is expanding its business from C-end applications like MeituXiuxiu to B-end services.

At the technological innovation front, the company's AI portrait generation effect has attained domestic supremacy, and the AIGC strategy is poised to unlock paid opportunities in the B-end market.

Currently, its productivity products, including Meitu Design Studio and Kaipai, offer continuous monthly subscriptions priced at 18-20 yuan per month, markedly higher than the 12 yuan per month for lifestyle and entertainment products like MeituXiuxiu and BeautyCam. The bank believes that with AI technology's support, the company's product paid penetration rate and ARPPU (average revenue per paying user) are expected to continue rising.

Subsequently, the launch of AI Agents will further augment product premium capabilities.

Huachuang Securities predicts that Meitu's operating revenue in 2025 will reach 4.5 billion yuan, a year-on-year increase of 26%. This optimistic outlook may partly stem from the growth momentum that AI Agents are poised to deliver. With AI technology's backing, the company's product paid penetration rate and ARPPU are anticipated to continue increasing.

Moreover, the overseas market represents another significant stage for AI Agents. Meitu's globalization strategy has yielded remarkable results: In January 2025, Meitu's overseas revenue ranked third among non-game vendors in China.

In February 2025, BeautyCam's "AI Outfit Change" feature became a hit in Thailand, sparking a buzz on Southeast Asian social media. Notably in Thailand, BeautyCam topped both the App Store and Google Play overall charts, and led the Photography and Video category on the App Store in 12 countries and regions.

As of December 31, 2024, Meitu's global monthly active user base stood at 266 million, a year-on-year increase of approximately 6.7%. Among them, the number of monthly active users in countries and regions outside mainland China was 94.51 million, a year-on-year surge of approximately 21.7%, accounting for 35.6% of the overall monthly active user base. International business has emerged as a crucial growth engine.

Regarding this, the head of technology marketing at Meitu Imaging Lab (MT Lab) stated: "2025 is a pivotal year for large model implementation. Our overall market has indeed shifted its focus from large models to 'large models +'."

This statement underscores the strategic positioning of Meitu's AI Agent—not as a large model per se, but as a focal point for the application and implementation of large models in vertical scenarios.

As AI technology iterations accelerate, Meitu's foray into AI Agents may herald a new trend: When technological dividends become ubiquitous, a company's success hinges not solely on algorithmic advancement but also on its engineering capability to translate technology into user experience, coupled with a profound understanding of vertical scenario needs.

This pragmatic approach of "emphasizing implementation over concepts" could be the key for mid-tier companies to stage a comeback in the AI era dominated by giants.