The significance of GPT Store may be underestimated

Text|Read finance and economics.

The GPT store can be understood as Apple's "App Store" in the field of large models. To put it simply, the value of GPT stores is that they give everyone more opportunities to create applications and services based on GPT technology, and you don't even need to know how to program, just through conversational chat, you can create an exclusive and personalized GPT.

Referring to the development of App Store, the launch of GPT Store has made it possible to build a new AI-based economic ecology in the future, and is expected to break the problem of AI commercialization.

But that's not the whole story. At a time when the AI industry chain has not yet matured, OpenAI's launch of GPT stores obviously has more significance.

On the one hand, the GPT store has made the relationship between the model layer and the application layer more delicate, and the previously sought-after shell AI application company may gradually die. On the other hand, in the GPT store, users can build agents similar to AI agents, further narrowing the distance with users. This will open up more business possibilities. So far, the biggest chapter in the story of the AI industry, the agent wave, has begun.

01 The "late" GPT storeGPT store first appeared at OpenAI's first developer conference in November last year. At the time, Sam Altman unveiled the product to everyone. The launch of the GPT store, which coincided with the public debut of ChatGPT, was originally scheduled to launch at the end of 2023, but was later postponed due to changes within the company. Nowadays, GPT stores are really coming.

Unlike the program store in the traditional sense, the GPT store is a no-** revolution. To put it simply, users don't even need to know how to code, just through conversational chat, to create an exclusive and personalized GPT, which can be designed for private use, or dedicated to internal use within the company, or earn money through the GPT store.

In the previous demo session, Altman himself created a startup mentor GPT in a few steps, Altman first opened the GPT builder (GPT builder), and first typed a paragraph to define this GPT, similar to helping startup founders think about business ideas and get advice.

Then, in the dialogue, the GPT Builder generates the name and icon of the GPT by itself, and asks Ultraman if he wants to adjust the generated name and icon by talking to him.

Next, GPT Builder asked him how the app interacted with users, and he said that he could choose appropriate and constructive answers from his past speeches, and then uploaded a paragraph of his previous speeches. After 3 minutes of creation, the person who accesses this GPT will receive a conversation beginning automatically generated by GPT, and can consult about the content related to the startup of this GPT conversation, and get an answer similar to Ultraman himself.

After the GPT store is released, the steps for users to use the GPT store are also very simple: the first step is to open GPT4** and click Explore;Step 2: Click on Creat A GPT;Step 3: Enter a name, description, and prompt wordsStep 4: Save GPT Personal Assistant. It is worth mentioning that the GPT store also launched 16 bots developed by OpenAI, including math tutors, creative writing coaches, assistant chefs, and more.

From the current point of view, the emergence of GPT stores has the opportunity to completely change the rules of the AI industry:

On the one hand, before the era of the "GPT Store", creating and deploying AI solutions was limited to people with significant programming knowledge and resources, and the GPT Store paved the way for widespread adoption of AI by lowering the barrier to entry and involving more people.

On the other hand, the emergence of GPT stores has also made it possible for artificial intelligence to build a new economic ecology. While the changes in AI are already amazing, finding ways to monetize them has become a challenge. The latest data shows that while 46% of SaaS companies launched AI capabilities in 2023, only 15% have found ways to monetize with this new feature.

The emergence of GPT stores has the potential to change that. This can be seen in 2008 when Apple released the App Store. In retrospect, the greatest value of the App Store is that it has created a new economic ecosystem: it not only provides a convenient and efficient software sales platform for third-party software providers, but also allows itself to obtain considerable commercial returns.

Of course, as the most concerned and influential company in the AI industry, OpenAI's impact on the release of the GPT store is far more than that.

02 The game between the model and the application layer continues, and the GPT store will have a great opportunity to become another "fire" to ignite the AI industry. But not everyone is happy about that. For example, AI application companies that are downstream customers of OpenAI may not be optimistic.

Today, the value of AI is undeniable. But there is one question that always lingers in everyone's mind: what should be the relationship between the application layer and the model layer?

Until the explosion of ChatGPT, this was not a problem at all. Even compared with OpenAI, application products based on OpenAI models are highly sought after by investors because of their clearer business logic.

In January 2022, Jasper's team of just nine people expanded to more than 160 in 10 months, with an estimated annual revenue of $60 million. In October 2022, Jasper raised 1$2.5 billion, reaching a $1.5 billion valuation just 18 months after its founding.

When products like Jasper and Snazzy launched in early 2021, the GPT-3 still needed high-precision prompts to report excellent results. This makes copywriting startups even more important because they bridge the gap between the sheer complexity of AI language models and the utilitarian needs of end users.

And just 1 month later, ChatGPT came. The question before the Jaspers became how to prove that they could hold on for the long term. At least for now, the situation is not promising. This summer, Jasper cut its full-year revenue forecast by 30% and began layoffs in July. According to The Information, Jasper has lowered its employee-facing ** valuation by 20%.

Such a change will not only occur in the application layer, but also in the middle layer, and some companies that do the middle layer will also have a hard time surviving. For example, LongChain reduces the difficulty of developing large model applications by encapsulating and packaging large model-related development components together. Now, OpenAI solves it with the Assistants API.

Now, with the launch of the GPT Store, it looks like the scales are tilting towards the model layer. This also seems to confirm that Sam Altman said at the YC alumni sharing meeting that OpenAI's model products will gradually expand, "In the case of increasingly limited living space, the so-called AI company of ChatGPT will die."

Since the model company will come to an end, is there no chance for the application layer?Probably not. At present, although there are many applications of large models, they are far from blooming, and even most of the application layers and model layers look thin, which will naturally be broken at a poke.

In the long run, the boundaries of large models have emerged, and applications built on large models are still in the early stage of exploration and are not yet clear. However, one thing is certain, the exploration of the AI application layer in the future will not only test developers' in-depth understanding of models and scenarios, as well as the AI-native application thinking that breaks the inertia of the Internet, but also pay more attention to the product power of product managers.

03 The possibility behind the agent of the new subjectIf we say that the game between the model layer and the application layer is just a continuation of the trend of the Internet industry. Then, another big innovation of the GPT store is the birth of the agent.

In the GPT store, users can build agents like AI agents and leave complex problems to it. AI agent is very different from our traditional understanding of robots, according to Bill Gates, there are three main differences between the two: proactively propose solutions based on user needs;Ability to complete tasks across applications;Improves over time.

According to this statement, we can even understand AI agent as a new subject. Following this logic, AI may give birth to new business models. In the past decade or so, the Internet has gone from portals to search to recommendations, essentially doing one thing: continuously improving the efficiency of information circulation.

But there are also problems with it, because the mechanism of the Internet is still label matching, running under a set of established rule systems, and in the face of many personalized needs, it still needs to be solved through manual customer service or other means, which affects the improvement of its conversion rate to a certain extent.

Compared with the tag matching mechanism, the possibility of AI agents is much greater. For example, if you like a certain taste of red wine from a certain production area, and then you also have a **comfort zone in your mind, AI Agent can choose wine for you around the world, and you don't even have to go through an intermediate platform, two AI Agents can complete the transaction.

In terms of work, AI agents can not only complete the work alone, but also find other AI agents to cooperate to form a new workflow and complete complex work together. If so, the GPT store can be the ** of the AI era, or it can be the UpWork (own career platform) of the AI era.

Although the future of the evolution of the AI era is not yet clear, it is certain that when intelligence is powerful enough to understand humans more and more, the product logic may also change greatly, such as from improving the human experience to improving the experience of the agent.

At the current pace of AI development, such a point may come sooner than we expected. In the future, when we revisit the starting point of the agent story, it may have started at the moment of the GPT store.