2024 Generative AI Application Ecosystem: Three Major Camps Emerge, Hot Areas and Unmet Needs

1 Internet giants, AI newcomers, and AI cross-industry players collaborate to build a general LLM ecosystem infrastructure

1. Generative AI large models are the next-generation general technology platform, enabling broad and generalized empowerment across industries based on foundation models and industry-specific fine-tuning.

2. As a new general technology, generative AI development still follows the S-curve pattern of innovation diffusion. The current stage sees continuous breakthroughs in model capability thresholds and scenario implementation gaps, indicating increasing technological maturity and market acceptance.

3. Compared to the CV visual recognition of the 1.0 era, generative AI has clearer and more diverse commercialization paths, attracting players from various fields.

4. Besides the early-entering internet giants, the emergence of AI newcomers is an inevitable result of the generative AI wave. Meanwhile, AI companies originally focused on B2B markets are also actively following suit. These three groups are jointly building the LLM general infrastructure ecosystem.

5. As an important C-end market, current large model companies' mobile traffic mainly comes from their own independent applications. Their own ecosystem traffic plays a crucial role in their AI applications.

For example, Ant Group's AI financial assistant Zhi Xiaobao (AI plugin) is supported by Alipay's massive user base; Douyin Group and Baidu's ecosystems back Doubao and Wenxin Yiyan respectively.

2 Intelligent agents may become new traffic distribution units, with embedded forms having differential advantages

1. Large models have shifted from technology-driven to ecosystem-driven. Commercial implementation can be broadly divided into software and hardware forms. The industry is expected to see significant upgrades or even restructuring.

C-end software forms can be further divided into "AI+" (native applications) and "+AI" (enhancement/empowerment of existing core businesses).

2. 2024 is considered the first year of AI phone development. Compared to previous generations of smartphones, AI phones focus more on personalized and scenario-based service capabilities while reconstructing the industry chain. They achieve intelligent agent distribution through user customization and manufacturer-specific tuning.

3. The integration and application of AI large models on different hardware devices further improve the intelligence level of devices, providing users with more convenient and intelligent services.

4. Taking intelligent vehicles as an example, as car companies successively launch city NOA advanced driving, the ability of AI large models to process sparse features during driving is expected to accelerate the development of advanced intelligent driving.

5. Recent strategic moves by large model companies in the intelligent agent field indicate that users may shift from the previous app usage model to interacting with intelligent agents. In this scenario, app stores as distribution channels may face ongoing business model challenges.

6. Compared to independent native applications, in-app plugin forms have potential differential advantages and a higher probability of breaking through in the competition within existing internet ecosystems.

7. Regardless of the carrier form, the fundamental purpose is to solve users' pain points. All services rely on the existing rich commercial service ecosystem, aiming to meet user needs in specific scenarios.

8. Independent AIGC applications have become important nodes for companies to attract customers and accumulate a large user base. At the same time, through AI plugins, companies are starting to delve into vertical fields such as education and learning, photography and beautification.

Although large model companies have limited presence in mobile social, mobile video, and financial management fields, their stable user base indicates huge development potential.

3 AI takes on multiple role positions, promoting the "APP+AI" trend across industries and stimulating AI transformation

1. Social entertainment, education and learning, and business office are high-frequency user scenarios. In summary, AI plays two major roles in these three scenarios: productivity tools and emotional value output.

These three scenarios have become popular applications, highly related to the focus of intelligent agents in current AI native applications.

2. Although life services are not listed in the TOP 3 scenarios, it's not difficult to see from user feedback that