Author: Frank, PANews

As AI moves from being a technological marvel to being practical, the rollout of AI applications is accelerating to meet growing consumer demands. Meanwhile, with the continuous improvement of large-scale modeling capabilities, AI seems to have entered an era where "everyone can create product prototypes."

During muShanghai AI Week, a roundtable discussion hosted by PANews, titled "Innovative Practices and Path Explorations in the AI ​​Consumer Ecosystem," focused on the real-world implementation paths of consumer-grade AI products. Participants included Feng Wen, Product Manager of the MiniMax Open Platform; Levy, CEO of FateTell; Anita, Head of Sentient APAC; and Gao Jiafeng, electronic musician and independent developer, representing diverse fields such as model open platforms, cultural export applications, open-source AI ecosystems, and music creation practices.

According to the guests, the core issues of consumer AI have not become simpler due to technological iterations. As model capabilities leap forward, the real barriers are shifting to scene understanding, data organization, user education, emotional value, and the construction of an open ecosystem.

AI has not made starting a business easier; the real barrier remains the application scenarios.

A common paradox in the AI ​​industry is that as models become more powerful, the barriers to entry for startups seem to decrease, but many products struggle to find sustainable application scenarios. Applications that seem viable today may quickly lose their relevance with the release of the next model version.

In Feng Wen's view, for consumer-grade AI products, product ideas and scenario judgments are still more important. As a provider of large models and open platforms, MiniMax will place more emphasis on the underlying model capabilities, token-related product design, and end-to-end experience for developers. However, from an entrepreneur's perspective, products should be designed according to "the intelligence level of the model six months later."

His assessment is that, given the continued validity of model scaling principles and the ongoing improvement in model capabilities, entrepreneurs should not be overly constrained by the current speed, cost, or capability limitations of their models. Instead, they should be more proactive in considering their target users, specific scenarios, and the problems to be solved. Model vendors will continue to provide cheaper, faster, and more cost-effective capabilities, while the application layer needs to more clearly answer the question, "Why this particular scenario?"

Levy added another source of barriers from the application layer. He believes that technology changes rapidly, but the data and understanding corresponding to a particular scenario will not be quickly erased. In the past, many people believed that only fine-tuning the model could create a data barrier; however, with the maturity of context engineering and prompt word engineering, the data and structures accumulated in context management can also change model performance. In particular, some highly vertical data, related to culture or personalized experiences, may not necessarily enter the weights of general models, which may become the basis for differentiation for consumer AI products to resist model iterations.

Anita offered a more cautious perspective on the idea that "AI lowers the barrier to entry for startups." She believes that while AI does make generating demo samples, building prototypes, and quickly launching an initial product easier, the truly difficult aspects of entrepreneurship haven't disappeared, and may even be more pronounced: customer acquisition, building community engagement, commercialization, and establishing connections between people beyond programming. She noted that the concepts of "super-individuals" and "one-person companies" are currently attracting much attention, but truly successful individuals often require more complex skills than simply calling upon large models.

From Bazi (Chinese astrology) to music: Understanding users better becomes a barrier to entry for consumer-grade AI.

As technological capabilities continue to advance, the value of consumer AI products will ultimately come back to human needs.

FateTell's practice provides a typical example. Levy explained that FateTell is an AI-powered Eastern fortune-telling/Bazi (Four Pillars of Destiny) consumer application for overseas users, currently with users in over 90 countries. The team initially avoided the pure efficiency tool direction, instead focusing on spiritual consumption and emotional value.

In his view, understanding one's own destiny and seeking explanation and comfort are fundamental psychological needs that exist across cultures and have long existed. AI has historically struggled to build trust in this context, but advancements in the capabilities of models like DeepSeek R1 have objectively helped users and investors understand the possibility that "large models can perform complex reasoning and explanations." The hurdle FateTell faces is not just model capabilities, but also how to translate and interpret Chinese cultural concepts such as the Heavenly Stems and Earthly Branches, the I Ching, and the Eight Characters (Ba Zi) for overseas users, and how to make these concepts accessible to people from different cultural backgrounds through language, visuals, and interaction.

Gao Jiafeng raised a similar question from the perspective of music creators: AI should not only deliver results, but also preserve the process. He mentioned that tools like Suno make music generation very direct, but they also skip the creative process, resulting in a lack of user participation and a sense of belonging. For musicians and ordinary users, creation is not just about getting a "finished song"; the process itself is part of the experience.

He used football as an analogy: even if ordinary people can never surpass Messi or Ronaldo, they will still play football because of their love for it. The same applies to music creation. Gao Jiafeng is developing MusicAIGameBoy (a music AI game console), which attempts to drive music code through large or small AI models, combined with gamified interaction, so that people who don't understand music can also participate in creation while playing. For him, the real scenario is not "automatically generating a song," but returning the interactive process of music creation to the user.

With the rise of agents, the logic of user education is changing.

In consumer AI products, user education often determines whether the product can actually be used.

Feng Wen mentioned that some users of the MiniMax open platform have basic development skills, but they are still hindered by API documentation, parameters, error codes, and token usage methods. To address this, the platform provides a model trial platform, development guides, demo cases, video tutorials, and other resources to help developers quickly move from understanding to application.

As agents evolve, the way users are educated is also changing. In the past, users needed to read documentation, understand interfaces, and troubleshoot errors. But with the performance upgrades of agents, many users now have the agent directly read documentation, search for solutions, select the appropriate model, and automatically correct the path. Model vendors need to improve the model, documentation, and platform experience, while the community, developers, and various product types will collectively lower the barrier to entry.

For Sentient, an open ecosystem is itself part of user education and product implementation. Anita explained that Sentient focuses on the open-source AI ecosystem and related infrastructure, and gathers developers through hackathons, grant programs, and other means. She emphasized that products must first clearly understand their target users: who are they, where do they appear, and through what channels do they build trust? For developer tools, hackathons and ecosystem collaborations are effective entry points; while for consumer products, KOLs, KOCs, and social media content are equally important.

With the rapid decline in AIGC costs, startups can produce trailers, visual materials, and promotional content at a lower cost, allowing their products to reach their first users more quickly. Gao Jiafeng also believes that product design should be as user-friendly as possible, allowing users to learn naturally through interaction and entertainment, rather than relying on extensive manuals. This "learning through use" approach may be more suitable for consumer AI than traditional tutorials.

As hardware enters the real world, the value of personalization and emotion continues to amplify.

Looking ahead three to five years, the guests generally believe that the AI ​​consumer market will still be in its early stages of penetration, but product forms will undergo significant changes.

Feng Wen predicts that smart hardware, robots, and embodied intelligence will reach a significant turning point in the next three to five years. With improved model capabilities, AI will no longer exist solely within software interfaces but will also enter the real physical world, performing more interactions and tasks. Some products will be geared towards humans, providing efficiency improvements or emotional value. Others may target agents, providing AI with the environment, tools, and infrastructure to connect to the physical world. Regardless of the form, products should ultimately remain human-centric, allowing people to spend more time on interpersonal connections, family, the real world, and richer life experiences.

Levy believes that making predictions over three to five years is already very difficult in the AI ​​industry, and even three to five months is fraught with uncertainty. He argues that while cutting-edge users are already heavily utilizing tools like ClaudeCode, most ordinary users are still in the early stages of AI adoption. In the coming years, AI will further meet more nuanced and personalized needs. Compared to the relatively "one-size-fits-all" services of the mobile internet era, AI has the opportunity to provide more specific and segmented services to everyone. At the same time, the anxiety and uncertainty surrounding unemployment brought about by technological development may further amplify the demand for emotional and psychological support services.

Anita summarizes this shift as "technology equality." She believes that in the future, the distinctions between humanities, sciences, arts, and technology will weaken. A small vendor might even improve their business by using AI to create advertisements and target information. The value of AI isn't necessarily about making everyone a top programmer, but about helping people in different life scenarios access better tools. At the same time, fear of unemployment and loneliness will drive up the demand for emotional value, creating more opportunities for hardware, AI pets, companion devices, and multi-sensory interactive products.

Gao Jiafeng, on the other hand, starts from the perspective of changing cultural forms. He believes that in the future, content formats such as music, film, and video will be reorganized, and it's even uncertain whether "song" will remain the smallest unit of music consumption. Current concepts like multi-track audio and audio tracks may continue to be broken down into more atomized creative units in the future. However, as forms are dissolved, the emotional connections carried by IPs, brands, and specific individuals will become more important. People are not always looking for perfect works, but rather for works with flaws, warmth, and the ability to build emotional relationships.

Although the guests did not provide a unified answer on consumer-grade AI, the discussions from different fields such as model platforms, cultural applications, open-source ecosystems, and music creation all pointed to the same trend: as model capabilities continue to improve, the competition in consumer-grade AI is no longer just about "who calls a stronger model," but about whether it can understand more specific users, real-world scenarios, and emotional needs.

The future AI consumer ecosystem may simultaneously include stronger open infrastructure, lower development barriers, more personalized services, more companionable hardware, and more new product forms centered around culture and the creative process. Models will continue to evolve, but what truly endures are those products that are needed, understood, and connect with people.