Not Just GPT Store, OpenAI Also Plans to Develop an AI Search Engine
-
Two months after releasing the GPTs (custom versions of ChatGPT) feature, OpenAI has finally launched the highly anticipated GPT Store today.
OpenAI claims that users have created over 3 million GPTs, with many creators sharing their GPTs for others to use. However, the GPT Store is currently only available for "paid access," open to ChatGPT Plus, Team, and Enterprise users.
Among these, ChatGPT Team is the latest paid version introduced by OpenAI today. When billed annually, ChatGPT Team costs $25 per user per month; when billed monthly, it costs $30 per user per month. GPT Store is considered the next App Store, and its commercialization has become a key concern for users. OpenAI stated it will launch the "GPT Builder Revenue Program." In the first phase, US developers will receive compensation based on user engagement with their GPTs. Over time, OpenAI will provide detailed information about payment criteria.
In the mobile internet era, controlling the gateway meant controlling the market. In the AI era, can OpenAI's GPT Store become that super traffic gateway?
ChatGPT is the fastest-growing application in history. In early January 2024, data from research firm Writerbuddy showed that among the top 50 popular AI tools globally, ChatGPT's traffic far outpaced others, ranking first. Among all AI chatbots, ChatGPT accounted for a staggering 76.31% of total traffic. With the official launch of GPT Store today, Google's worst fear might be coming true—ChatGPT's massive user base could potentially rewrite the rules of internet traffic distribution through GPT Store.
The most significant difference from App Store's product shelf logic is that GPT Store prominently features a large search box upon opening, resembling more of an "AI search engine." Users can directly input keywords here, with support for Chinese language searches.
Below the AI search box lies GPT Store's categorized product shelves: including Featured Picks, DALL-E Image Generation, Writing, Content Creation, Research & Analysis, Programming, Education, Lifestyle, and other categories. GPTs are currently ranked by popularity, meaning the more users a GPT has, the more likely it is to be recommended. New GPTs are added every week. The officially recommended popular GPTs are:
Developer websites are attached to the names of GPTs, allowing direct navigation, which will become a significant traffic source. The more users a GPT has, the greater the exposure for the associated website.
Additionally, in terms of search results, GPTs significantly improve search accuracy compared to traditional search engines. For example, when using AllTrails to request "easy hiking trails near New York City" (a preset prompt), it provides a detailed, image-rich trail plan and direct links to third-party websites. The earlier GPTs are developed and the more users they attract, the greater the early traffic dividends they will reap, creating significant appeal for developers.
Perhaps, GPT Store will directly challenge traditional search engines. The future internet search gateway may gradually shift from search engines to generative AI conversational entry points like GPT Store.
In seizing the traffic portal of the AI era, OpenAI has finally taken its first step. In the mobile internet era, the App Store helped Apple build a strong moat and maintain its position as the top tech company for a very long time. Will GPT Store be the next App Store?
Li Guanghua, co-founder of LanguageX, told "Jiazi Guangnian" that his attitude towards GPT Store has changed over the past two months.
When OpenAI first announced the launch of GPT Store in November last year, Li Guanghua believed that GPT was just a large language model, lacking the infrastructure dominance of iPhone and iOS. He thought it might only capture a certain share among large model companies, far from matching Apple's App Store. In practice, the actual results were also unsatisfactory, with poor customization effects. But today, Li Guanghua's thinking has changed. He believes that GPT Store 1.0 might resemble an App Store-like marketplace, but version 2.0 could introduce entirely different dynamics and possibilities—not just more applications, but a future 'AI model marketplace'. Even more astonishing is the idea that interactions and transactions between GPTs or Agents could be spontaneous, with GPTs buying and selling services among themselves.
In this scenario, GPT Store would no longer be just a tool marketplace, but a 'silicon-based talent (Agent) market'. Agents could independently complete tasks, proactively seek collaborations with other Agents when needed, and combine to form new workflows to handle complex operations. If this vision holds, GPT Store would become the AI equivalent of a freelance platform, with OpenAI emerging as the world's largest silicon-based talent company—much like how Airbnb is essentially the largest hotel chain globally, and Didi operates as the largest taxi company. After watching OpenAI's Dev Day last November, Dai Yusen, Managing Partner of ZhenFund, shared his perspective—OpenAI resembles a combination of 'Intel+Microsoft' from the PC era. It provides chips, an operating system, and several killer applications, creating barriers through technological upgrades and the economies of scale enabled by massive capital investment. Simultaneously, it acts as an operating system, app store, and super app.
Once the operating system of the AI era is established, the second-place player will have little chance to compete. Dai Yusen believes that while the economies of scale from large capital investments might still face challenges, the network effects and high migration costs of an operating system are nearly insurmountable, allowing the winner to take all. "After Dev Day, many lamented that OpenAI wiped out numerous startups. Essentially, this is because these startups were working on calculators, notepads, or browsers—either too simple with no barriers or tasks the operating system would inevitably handle."
Bill Gates predicted last November that AI Agents would be the next platform after large models, not only changing how everyone interacts with computers but also transforming our lives within five years. 2024 is expected to be the year of the AI Agent explosion. Therefore, apart from OpenAI, an increasing number of large model companies and AI application-layer companies are vying for entry points similar to the 'GPT Store,' namely Agent platforms. According to statistics, many global tech giants, domestic leaders, and AI startups have launched their own Agent platforms. In vertical scenarios at the AI application layer, software companies like Feishu and DingTalk have also introduced their intelligent partners or super assistants, which are essentially Agents tailored for business scenarios.
OpenAI's GPT Store is an Agent platform targeting the to C market, while in China, more companies are opting for the to B approach. Following the launch of DingTalk's Super Assistant this week, DingTalk's president Ye Jun explained to media outlets like 'Jiazi Guangnian' the differences between DingTalk's AI Agent Store and OpenAI's approach. Ye Jun stated: "The biggest difference lies in the scenarios. Over the past eight years, DingTalk has accumulated the richest scenarios from organizational digitization to business digitization, personalized data accumulation, and an executable system built with low-code. OpenAI has to build everything from scratch."
Zhou Hongyi, founder of 360 Group, has also emphasized the importance of scenarios. He believes GPTs are still far from becoming an AI App Store. The reason OpenAI released GPTs is to hope people will help them discover various application scenarios.
In the B2B market, there are many landing scenarios for AI Agents. AI practitioner Yang Chang summarized them into three categories: "First is sales, customer service, and marketing - areas where AI Agents can significantly improve efficiency and effectiveness. Almost all platforms focus on these areas. Second is low-code, no-code, and RPA - important assistants in the AI Agent implementation process. Before fully automated AI Agents arrive, these three offer relatively lower costs and more controllable results. Third is expert knowledge and SOPs - the key to successful AI Agent implementation in enterprises." He also mentioned that AI Agents have three implementation forms, which can coexist within a single company, including Copilot work assistants, business auto-piloting, and autonomous agents.
Currently, the implementation of AI Agents in the to B industry is still in its early stages, but the competition for ecosystem dominance and market entry has officially begun.