Change to deal cards with another person, and the agent of Baidu (06608) has also been present.

The capital market has a classic experience with K-line charts: whenever a company announces changes in its senior management, sell orders on trading software queue up and flood in. A large sell order in the morning sets the tone for continuous declines in the following trading days. It's as if everyone is selling a theater troupe that relies on one or two key figures to survive. This is not a rational judgment, but a conditioned reflex - seeing the word "change" immediately brings thoughts of "risk". The capital market is never about trading companies, but about trading people's perception of companies. And this "perception" is wrong 90% of the time. So, 10% of the people, using their perception, make money off the other 90%. The rules of the capital market are actually quite simple. Whether a company is worth holding or not depends on whether its products are selling, if customers are staying, and if the business model can sustain itself. As for whether a change in the announcement board will affect the above three points? It can, but often this effect is magnified by the K-line chart. A stock price adjustment does not mean the company is deteriorating, sometimes it's just the shareholders panicking. Recently, there has been concern about Baorong Intelligence (06608). What the market truly cares about is not where a particular person is going, but a more fundamental question: Is the production line that can create a silicon-based employee in two weeks still operational? Will the more than 200,000 VoiceAgents and silicon experts working at customer sites go on strike if someone leaves? Will the 98% customer retention rate plummet because of a unfamiliar signature on an announcement? If the answer to these three questions is no, then every moment of panic in the recent market is simply handing over chips to those who understand. Money in the capital market always flows from the hands of the panicked to the pockets of the calm. There is a law in business history that has been repeatedly verified: a company propped up by a single individual never grows big; a company that can sustain itself relies on a system, not a savior. When Microsoft announced Bill Gates' complete withdrawal from daily management in 2008, the market reacted similarly to today - with concern, doubt, and stock price pressure. But as everyone later found out, after Gates left, Microsoft's market value increased from less than 300 billion to 3 trillion. It's not that Gates wasn't important, but rather, before he left, he had transformed Microsoft from a "founder-driven company" to a machine that could run on its own system. The strength of Windows, the stickiness of enterprise customers, the commercial model for software licensing - these three pillars were stable, regardless of who became CEO. The same logic can be applied to Baorong Intelligence's current situation. A company's lifeline is not hanging on the office door of a particular person, but engrained in its products and business model. Looking at the products first, over 60% of Baorong's revenue is related to the VoiceAgent product. The peak call volume last year reached 100 million calls, with a completion rate of 81%, compared to the industry average of 65%-70%. This ten-point difference in completion rate translates into business language as: for the same 100 AI voice calls, Baorong's Agent can complete ten more tasks than the industry average. The key difference between Agent and the previous generation voice solution, Siasun Robot & Automation, is the architecture - the traditional approach is keyword matching, while Baorong's Agent uses an end-to-end large model with a "bad case return-evaluation-emprovement write-back" closed loop mechanism, automatically returning underperforming calls for training, the system finds and addresses problems on its own, and can even rollback if necessary. This means that the Agent is not just a static software delivery, but a silicon-based workforce. Another comparison reveals that a similar product from a major internet company has a completion rate of 73% in the same testing environment, lagging by 8 points. The difference is not in model parameters, but in the accumulation of vertical scene data. Baorong's VoiceAgent has been running in institutional customer applications for eleven years, feeding dialogue data from 8,000 customers, naturally understanding when to probe further and how to maintain a friendly tone. This barrier is not related to technology, but to time. This product capability is not an asset locked in the mind of a technical expert, but rather solidified in the system capabilities of the "result cloud". The result cloud is divided into three layers: the base layer "Baiji" is computational infrastructure with exclusive AI models, the middle layer "Baigong" is an intelligent operating system, and the upper layer "Baihui" is an "application store" for silicon-based employees. This three-layer architecture encapsulates the large model, inference engine, task configuration, and scene connection into a standardized pipeline. In other words, the pipeline is Baorong's "AI production capacity," so it doesn't rely on any one person managing it. As long as the production base for Agent is still operational, a new silicon-based employee can be created in two weeks and deployed on-site to work directly for the client. This is comparable to Ford's assembly line for manufacturing cars - a handcrafted workshop stops if the master worker leaves, but a production line will continue to output regardless of who mans it. Baorong's current moat isn't a person, but the pipeline itself. . Baorong's MaaS business surpassed 1 billion in revenue by 2025, growing by 9% year-on-year, with a core customer retention rate as high as 98%. While B-end customers inherently lack the preference to switch AI service providers frequently, with an average revenue per customer of 3.59 million, which has increased by 6% year-on-year, indicating that existing customers not only remained but continued to increase their purchases. Customers are the most honest people; they don't vote with their words, they vote with their renewals. A 98% renewal rate actually means that Baorong's AI capability has already been bonded to its customers. Because in the pursuit of cost-effectiveness and efficiency, human nature is amplified within the enterprise - once a silicon-based employee like VoiceAgent is utilized, it's impossible to return to the stone age of "press buttons to reach a live agent". Regarding the business model, Baorong has undergone three iterations in its eleven years of operation: from MaaS charging by API call volume, to BaaS sharing profits based on transaction size, to the formal introduction of RaaS by the end of 2025, directly packaging AI as a "silicon-based employee", charging by position and based on results. If selling a product, it is naturally subject to comparison, but if the focus is shifted to results, future profit margins will increase, corresponding to using a new EPS to give a PE. , a company like Baorong, with AI product capabilities solidified in the "result cloud" pipeline, customer stickiness anchored by a 98% retention rate, and a business model locked into the benefits community of RaaS. As for who sits behind the desk in the office, that's a separate matter from the fundamentals. The capital market has a perennial flaw: treating personnel changes as fundamental changes to trade. Little do they know, the real fundamentals are never listed in an announcement, but are ingrained in the products, customer retention rates, and business models. In the 1980s, when Intel transitioned from memory chips to microprocessors, Andy Grove asked Gordon Moore, "If we were kicked out, what would the new CEO do?" Moore's response was to exit the memory business. Grove then questioned why they couldn't do it themselves. Intel then laid off a third of its workforce, closed seven factories, and completed one of the most famous strategic pivots in history. At the time, the market was in a panic, but later it was known as a "brave decision". If it's a good company, then the pain of transformation is actually a good opportunity to pick up bargains; when a good company's stock falls, pick it up. The difference lies in whether you are the one selling at the time, or the one regretting afterwards. A company's transition from "selling tools" to "selling Agent labor" will inevitably require a round of adjustments in organizational structure and talent. Ultimately, what the market should be focusing on is the strategic line and business continuity. As long as the pipeline of the result cloud is running, VoiceAgent is still answering calls, over two hundred thousand silicon-based employees are still working at customer sites, 98% of customers are still renewing, and the revenue-sharing model of RaaS is still operational. As long as these facts remain, every emotional pullback in the market is a misreading of the fundamentals. At the very least, the PE-Band provides a lower limit for price protection. The capital market has a particularly fair aspect: in the short term, it's a voting machine; in the long term, it's a weighing machine. During the voting phase, emotions determine prices; during the weighing phase, facts determine value. Right now, Baorong is in a period of transition from a voting machine to a weighing machine. When the number of silicon-based employees goes from two hundred thousand to fifty thousand, and when the RaaS revenue-sharing model goes from a benchmark case to an industry standard, the linear extrapolation of EPS at that time will bring a different PE valuation system. Because people tend to think that tomorrow will be worse in adversity, but feel that tomorrow will be better in advantages. Whether a company is worth holding or not is never about who's name is being swapped on the announcement board. It's about whether the pipeline is still running, if customers are still renewing, and if the business model is still evolving. If those three elements remain, everything else is just noise. And the biggest characteristic of noise is that it's loud, but not important. Perhaps Baorong Intelligence is currently at this stage.