The commercialization of large AI models is entering a new phase. As the "model war" cools down, companies are focusing more on the practical application effects and commercial value of large models, rather than blindly pursuing scale and quantity.
Currently, large model enterprises mainly have two commercialization paths:
-
Create high-quality general-purpose large models, which can both empower other enterprises as an open platform and develop into "super applications", following a technology + product dual-driven approach.
-
Transform from general-purpose large models to application companies, finding survival space in vertical domains.
However, regardless of the chosen path, they face challenges of high computing power costs and customer acquisition costs. Against this background, AI plugins have become a good entry point:
- Plugins are the lowest threshold and easiest-to-use form of AI-native applications.
- They can quickly obtain market and user feedback, reducing trial and error costs.
- Browser plugins can cover multiple usage scenarios, facilitating rapid user aggregation.
Currently, several companies including Doubao and Kimi have launched browser plugins with different functional focuses:
- Doubao's plugin is feature-rich, including search result organization, video content summarization, and text-to-image generation.
- Kimi's plugin focuses on lightweight search, offering highlight-and-question and summary features.
However, competition in the plugin field will also intensify. To succeed in this track, the key lies in providing quality user experience and powerful model capabilities.
Overall, AI plugins provide a good starting point for large model commercialization exploration, but commercialization remains a long-term challenge. Enterprises need to continuously exert efforts in product experience and technological capabilities to stand out in the fierce market competition.