She Zongming/Wen After launching the paid subscription function, Doubao has taken another step forward on the commercialization path by charging channel service fees.

In recent days, the saying 'Recommended hotels for tofu buns will be charged' has become popular. The incident originated in late July when Doubao announced that it would officially implement new regulations on specific channel service fees starting from August 10th. This is the first commercial case in China to implement large-scale transaction commission collection through AI (artificial intelligence) dialogue entry.

After this sparked heated discussions, the public relations manager of Doubao publicly responded that Doubao currently does not have paid promotion and only charges channel service fees, which are only deducted after the order is completed,AndThe dual charging model of "commission+advertising" is different. Since then, Tiktok has extended the effective date of the new regulations to August 20.

There is no forever free lunch in the field of AI. Doubao's "commission" is the conversion button that has to be pressed under the pressure of reality.

To this day, Doubao has over 300 million monthly active users and billions of daily active users. However, on the other hand, as a national level application, there is a cost black hole - according to industry estimates, Doubao's daily computing power cost may reach tens of millions of yuan, but its daily income is less than one million yuan, resulting in a serious imbalance between income and expenditure. In the context of the unsustainable trend of "burning money for scale", finding a profit model is urgent for Doubao.

Given that the top domestic models have not yet broken through the bottleneck of "basic free+high-end subscription", the penetration rate of C-end members is limited, and the overseas advertising model is also not adapted to the domestic environment, it is reasonable to focus on "increasing the amount of transactions within the ecosystem" for monetization. Relying on Tiktok's huge local life business system, Doubao has completed the transformation from a tool assistant to a trading portal, and opened the whole link of "AI consulting - demand matching - precise recommendation - closed-loop trading". It is easy to draw commission.

Not long ago, Liang Rubo, CEO of ByteDance, said at the staff meeting that he hoped that Doubao would become a parallel backbone business with Tiktok in the future and incubate independent branches such as e-commerce and life services. Before that, Doubao acted as a super portal inside the byte - it connected search, content and local life, and led the user's intention back to Tiktok. Now, it is clear that Doubao has been designated as an independent trading channel, and an exclusive rate system has been established to make Doubao "de Tiktok diversion tool".

As one of the largest AI native applications in China, Doubao's clear pricing of AI traffic will inevitably have multiple impacts on local life service merchants and platform ecosystems.

The core concerns of merchants are focused on two points: firstly, whether AI entry can bring real new customer sources still needs to be verified, but the cost of new channels already exists. Take the hotel as an example, the comprehensive rate of Doubao channel is 12%, which is lower than the traditional OTA (online travel platform) common 15% -20%, but 4 percentage points higher than the 8% of the same category of Tiktok main station.

The underlying logic of this design is that AI recommendation belongs to the precise intention traffic initiated by users, and its conversion efficiency, matching accuracy, and transaction certainty are theoretically higher than blind investment of information flow. However, it is still unknown whether more expensive precise traffic can bring higher returns to merchants and whether it will become a re pricing of existing traffic.

In addition, if the merchant does not want to bear this service fee, can they close the recommendation in this scenario and refuse this fee? The response of Tiktok's official customer service is that this function currently does not support merchants to shut down on their own, and the costs incurred in this scenario cannot be chosen by the merchants to not bear, and the merchants can apply for exit according to the process. Merchants may have doubts about the convenience of exiting channels.

Secondly, although the official response from Doubao states that there is "no paid promotion" and "no advertising fee", the platform is bound to face severe challenges in balancing algorithm neutrality and commercial monetization.

Unlike traditional search engines that provide multiple links for users to filter independently, AI conversations often only offer a small number or even a single option, and the commercial value of "AI priority recommendation" is self-evident. When the AI recommendation ranking logic is directly linked to the platform's commercial interests, merchants naturally worry that the AI recommendation weight will quietly tilt towards high commission formats and specific enterprises - especially in the case of vague AI recommendation rules.

In addition to facing the above doubts, Doubao also needs to accept users' trust questioning about whether the AI recommendation results are reliable. In fact, previous incidents such as "netizens using bean buns to book restaurants, only to be rejected when they arrive" have left them with the impression of "having more emotions than practicality".

Although the above issues involve multiple levels, the core points to 'how to gain trust from others'. This tests the ability of Doubao to seek balance in multi-party games: it needs to cover both the cost of technology investment and the survival space of merchants, while also safeguarding the bottom line of consumer rights. How to ensure transparency of order sources and deduction details, interpretability and supervision of AI recommendation mechanisms, and clarity of rights and responsibilities. These are all mandatory questions that cannot be avoided.

Bean bun commission is a landmark event in the reevaluation of AI entry value. From cases such as Qianwen binding to Taobao for CPS (based on actual sales revenue) shopping recommendations and embedding Yuanbao into the JD ecosystem, it can be seen that the "dialogue recommendation order sharing" model is becoming an important direction for the commercialization exploration of AI assistants. The dividend of free AI traffic is destined to fade away, and the era of "toll stations" for AI entry is coming. The standardized AI commission rate for Doubao is expected to provide pricing reference for the industry in the process of testing market trends.

But the trend is the trend, and AI assistants still need to make their commission payments convincing - they need to ensure that the pricing system that is compatible with the native traffic value of AI is not a "confused account", and the fairness and rationality of AI recommendation rules will not be eroded by algorithmic black boxes. Otherwise, the recent barrage directed at the platform by Zhong Suisui, who used algorithms to adjust commission rates for orders and intervene in traffic distribution, as well as opaque fees, could easily fall on the top of the AI super entrance.

(The author is a senior media professional)