Following its separation from Alibaba Cloud, Tongyi To C remains uncertain about its direction for 2025.

A visit to Alibaba's recruitment website reveals details of Tongyi To C's hiring activities. In the "Intelligent Information" segment, six positions related to "Tongyi" were listed, with their initial release coinciding with news of Tongyi To C's spin-off last year:

As part of the Intelligent Information Business Group, Tongyi's To C team continues to recruit new product managers, front-end developers, and engineers. However, the number and types of positions, along with their job descriptions, suggest that Tongyi To C is still reliant on its path at Alibaba Cloud, aiming to create hits through technological prowess.

Is this approach viable for an independent Tongyi To C? Despite its access to leading domestic technology, considering its performance over the past year, this path may not be suitable.

Tongyi To C should consider focusing on building a C-end ecosystem and prioritizing user growth. This raises the question of whether to emulate Doubao's fast-paced, simultaneous approach to doing and thinking, or Quark's method of building an ecosystem from a specific scenario and integrating AI capabilities.

01. Tongyi To C's Technology-Oriented Path Dependence

Current job postings in the Intelligent Information Business segment reveal two main characteristics of Tongyi: a lack of specific functional direction and a persistence in technology-oriented thinking.

In terms of functions, Tongyi To C has only one product manager position titled "Product Manager." In contrast, Quark, also in the Intelligent Information Business segment, has product manager positions with specialized titles such as "AI Search," "AI Image Vision Direction," "Agent," and "AI Creativity/Consumption." This suggests that Tongyi's To C exploration is still in its early stages.

Comparing Tongyi with Doubao, a contemporary app, Doubao's product manager positions are titled "Model Evaluation," "Security Policy," and "User Growth." This indicates that Tongyi's To C team may currently be in a state of "leaderless chaos," awaiting a overarching product manager to set the framework and plan specialized functions.

The job description of Tongyi's current sole product manager emphasizes a technology-oriented approach, focusing on "working closely with the algorithm team to transform algorithmic capabilities into product features." In contrast, Doubao and Quark prioritize user research and user-centric thinking in their job descriptions.

Undeniably, Tongyi To C possesses impressive technological capabilities. However, the question arises whether this technology-oriented mindset needs reevaluation in 2025.

02. Alibaba's Unsuitability for a Technology-Oriented C-end Approach

While there are successful cases of technology-oriented C-end applications, such as OpenAI, their path is almost unreplicable for domestic vendors.

OpenAI's structure, with parallel technology and commercialization teams, is typical of a To B Maas model. Although its structure has evolved over time, it remains focused on profitability. Such a technology-oriented structure is ideal for many domestic large model vendors.

However, in the domestic context, without suitable C-end product features and ecosystems, a technology-oriented approach will fail to gain C-end influence.

Over the past year, Tongyi app has explored the path of creating a C-end ecosystem through technology. However, it's evident that pure technical superiority can be overshadowed by product feature combinations, as seen in Doubao and Kimi.

Moreover, as previously discussed, technology is not currently Tongyi To C's biggest strength. Alibaba Cloud supports multiple large models, and Tongyi's To B business performance has little impact on the To C team.

Additionally, Alibaba's system does not have a highly consistent layout for its internal AI ecosystem, and each board is self-sufficient. Therefore, Alibaba Cloud does not rely on Tongyi To C's influence to drive sales.

03. Quark Path or Doubao Path

In the realm of large models, new players like DeepSeek-V3 have achieved remarkable results with limited resources. Its training consumes only one-tenth of the computing power of Llama-3.1-405B, earning it the nickname of "Pinduoduo" in the large model world.

Apps like Doubao, Kimi, and DeepSeek demonstrate that in resource-limited environments, large model vendors adopt different, more decisive approaches. In contrast, Tongyi, with seemingly comprehensive resources, struggles to enter the To C market.

As Alibaba's primary large model brand, Tongyi To C has access to support as long as it presents viable ideas. Therefore, it can learn from Quark and Doubao, the top two AI applications in terms of downloads in 2024.

Quark's approach involves building a C-end ecosystem from a specific scenario and integrating AI features to enhance it. Doubao, on the other hand, serves as a driver for ByteDance's cloud business, quickly establishing a C-end platform and connecting with other ByteDance ecosystems.

For Tongyi, Quark's significance lies in its ecological niche and growth path within Alibaba's system, while Doubao's relevance stems from its speed and user-centric thinking.

Ultimately, the product managers at Tongyi To C must personally discern the most suitable path based on this analytical framework.

*The lead image and images in the article are sourced from the internet.