Discussions surrounding the Token economy have intensified in capital markets since 2025. Computing power Tokens, model Tokens, and call Tokens all point to the same trend: AI is transitioning from a technological concept to becoming quantifiable, settleable, and commercialized. However, from the perspective of industrial evolution, a more critical distinction is emerging: computing power Tokens and model Tokens are closer to the supply of underlying capabilities, whereas scenario Tokens are closer to the final realization of value.

A "scenario Token" can perhaps be understood as a value unit within an AI-native application, where underlying computing power and model calls are translated into quantifiable business outcomes in specific operational contexts. What businesses are truly willing to pay for is not the underlying volume of Tokens consumed, but whether these Tokens ultimately deliver results—such as higher conversion rates, faster deal progression, lower customer service costs, and improved organizational efficiency.

Using a more intuitive analogy for the Token economy industry chain: computing power Tokens are like electricity bills, model Tokens are like fuel costs, while scenario Tokens are more akin to the "fare for getting a person to their destination." Corporate clients do not pay extra for "how much electricity was used or how much fuel was burned," but they will pay for the final outcome.

Consequently, the capital market's focus on the next phase of profit centers within the AI industry chain is gradually shifting from "who possesses more foundational Tokens" to "who is closer to the scenario outcome and who is more capable of defining scenario value." From this viewpoint, MARKETINGFORCE (02556) deserves re-examination.

This company initially started with marketing cloud and sales cloud platforms, gradually expanding into various operational scenarios such as customer service, business analysis, foreign trade, manufacturing, and government affairs. Essentially, it is following a path of upgrading from a "tool provider" to an "AI-native application platform." In other words, it sells not the underlying capability itself, but a system that processes that capability into operational results.

The 2025 performance figures demonstrate that this path is not merely conceptual. The company reported total revenue of RMB 2.818 billion, a year-on-year increase of 80.8%. Revenue from AI application business reached RMB 1.487 billion, up 76.5% year-on-year, becoming the largest revenue source. Adjusted net profit was RMB 152 million, surging 91.3% year-on-year. The number of Key Account (KA) clients reached 1,609, an increase of 105.5%, while the Average Contract Value (ACV) for KA clients rose by 60.6%. These figures reflect not merely scale expansion, but simultaneous improvement in high-value client numbers, revenue per client, and operational efficiency.

Concerning AI-native applications, a common market question is whether "general-purpose Agents will subsume vertical applications." This judgment overlooks two key points. First, commercial objectives are not aligned. Model providers naturally hope for higher Token consumption, while application providers are more concerned with achieving better results at lower cost. Second, the challenge in enterprise-level scenarios has never been solely about generation capability, but involves cross-system calls, process orchestration, permission governance, result feedback, and collaborative execution. General-purpose Agents can handle single-point tasks but are unlikely to replace enterprise-level systems deeply integrated into CRM, ERP, customer service, and business analysis platforms.

From this perspective, MARKETINGFORCE's competitive barrier lies not in model parameters, but in the combination of "industry know-how + knowledge graph + Agent middleware + engineering capabilities." The company has served over 210,000 enterprise clients cumulatively, covering 721 sub-sectors, and has accumulated thousands of reusable knowledge graphs across six core industries. The deployment cycle for new client-specific Agents has been shortened from three months to under three weeks. The knowledge middleware is responsible for accumulating and organizing industry knowledge, while the Agent middleware handles governance, multi-model scheduling, and multi-Agent collaboration. This indicates its platform capability has progressed from "being able to create demos" to a stage of being "deployable, manageable, and replicable."

More convincingly, MARKETINGFORCE not only sells AI to its clients but has also implemented AI within its own operations first. In 2025, the company's total headcount was 1,737, up 11.1% year-on-year, while overall employee productivity improved by 62.7%. The management expense ratio dropped to 6.8%, the sales expense ratio fell to 14.5%, and the R&D expense ratio remained at 15.6%. This indicates that growth was not achieved by cutting investments, but through the deep embedding of AI into internal processes like lead management, training, code generation, and financial dashboards, leading to genuine efficiency gains under sustained R&D investment.

In public communications, it has been mentioned that AI Agents, unlike traditional tools, are "natural digital employees," driving a shift in business from "people seeking digital processes" to "digital processes seeking people," thereby reshaping business logic. This statement deserves careful consideration by the capital market: when AI ceases to be just an add-on feature and begins to directly integrate into business workflows and operational decision-making, what clients are purchasing is no longer merely a software module, but a system of digital employees that continuously creates results and iteratively upgrades.

This is also why strategic cooperation deserves particular attention. For instance, a machinery equipment group visited MARKETINGFORCE's Shanghai headquarters for research and discussions, focusing on topics such as AI Agent ecosystem development, enterprise-level AI middleware, and digital-intelligent transformation in manufacturing. For the capital market, the significance of such cooperation extends beyond being "contract news"; it is a signal that MARKETINGFORCE is advancing from marketing and sales scenarios into the complex processes of physical manufacturing, promoting AI's evolution from a corporate growth tool to an industrial upgrade tool.

Looking further ahead, internationalization and ecosystem synergy are opening a second growth curve for the company. In 2025, revenue from foreign trade business reached RMB 75.934 million, a surge of 134.4% year-on-year. The platform supports 49 multilingual options, with reach exceeding 95% of global countries and regions. Overseas hubs have been established in Hong Kong (China), the United States, and Singapore. Simultaneously, the number of partners increased to 295, with ecosystem-contributed revenue growing 35.0% year-on-year. This implies that MARKETINGFORCE's growth logic is not solely reliant on its main business growth, but is a combination of "core business growth + self-use validation + ecosystem synergy + international expansion."

Ultimately, scenario Tokens warrant greater attention because they are closer to the outcomes for which businesses are truly willing to pay. What makes MARKETINGFORCE worthy of ongoing tracking by the capital market is not just whether it is riding the AI wave, but its potential to become a beneficiary of the value migration occurring at the application layer. If the market continues to understand and value it within the framework of a traditional software company, it may be underestimating not only its current profits but also its medium- to long-term value as an AI-native application platform.