Master LLMs with our FREE course in collaboration with Activeloop & Intel Disruptor Initiative. Join now!

Publication

The GPT Store: Is the Hype Justified?
Artificial Intelligence   Latest   Machine Learning

The GPT Store: Is the Hype Justified?

Last Updated on January 12, 2024 by Editorial Team

Author(s): Jorge Alcántara Barroso

Originally published on Towards AI.

OpenAI has just launched the GPT Store, which is now available to ChatGPT Plus, Team (new!), and Enterprise users. The store offers a space where developers can create and distribute their custom GPT-based chatbots. At launch, it features over 3 million custom GPTs across various categories.

The GPT Store is envisioned as a marketplace of generative chatbots, where each bot is an instance of OpenAI’s GPT-4 model, customized with specific prompts (instructions), a set of tools (retrieval, dalle, browsing, code interpreter & custom actions), and the ability to load files to add custom knowledge to use with retrieval. If you want to learn how to use these, check out this article from last year.

This customization allows these chatbots to focus on more particular tasks tailored to user needs. For example, a GPT could be loaded with a cookbook collection to assist with recipe queries or ingested with a given codebase to aid developers in code generation. A more advanced GPT may connect with external APIs (web applications) to take action for you.

The launch of OpenAI’s GPT Store has been met with a mixture of enthusiasm and skepticism within the AI community. Optimists see it as a milestone in AI’s evolution, a step towards more accessible and versatile AI applications. Critics, however, voice concerns about its potential impacts on the AI development ecosystem, including the implications for creators’ revenue and the challenges of quality control and ethical standards.

It’s essential to realistically assess the challenges & opportunities it presents.

The Good

With the upcoming builder revenue program, US builders will be able to earn based on user engagement, but expectations should be tempered as the revenue share is likely to be modest.

From a technical standpoint, creating these custom GPTs is accessible to a wide range of users. The process involves using OpenAI’s GPT Builder, where creators can simply describe in plain language the capabilities they want their GPT to have. This tool then attempts to create an AI-powered chatbot based on these specifications. The ease of creation has been a point of interest since its announcement, allowing for quick development and sharing of GPTs​.

The Bad

However, there are concerns among developers about the potential for copycat apps, reminiscent of issues faced in other app stores like Apple’s App Store in its early days. Furthermore, the moderation of these GPTs is nonexistent, which may lead to unexpected, undesired behavior or even custom actions that send your conversations or other information to 3rd parties without your explicit consent.

But, how do I make money in the GPT Store?

The Ugly

With 180 million ChatGPT users and 250k Plus subscribers [source], the market seems vast. But, let’s crunch the numbers. Assuming a 10% revenue share, the theoretical upper limit for all creators if they reach every Plus user is $6 million annually. A Plausible Scenario of a Successful GPT:
– 1% of Plus users engage with one of your GPTs.
– These users are equally engaged with 5 GPTs from different creators.

OpenAI earnings per Plus user: $20 / mo
Annualized Plus revenue: $20 * 250,000 * 12 = $60 mill
Revenue share for all creators: 10% of $60 mill = $6 mill
Your final share as one of five creators used by 1% of Plus users:

$6 million / (100 * 5) = $12,000 Annually

For a rev share of 20% and a doubling of Plus users, this can get to 48k/yr, which, compared to other 2-sided marketplaces, is quite insignificant even for the biggest winners.

Why is everyone talking about the store then? Let’s delve into the implications for OpenAI and developers.

Strategic Goals of OpenAI with the GPT Store

The GPT Store is not designed as a revenue source for creators, it does not create new engagement for OpenAI tools as it’s limited to Plus+ users. So, it must be a discovery tool, one that can teach OpenAI what to build next for its users. Building successful GPTs means teaching OpenAI where they should put their product development focus for their next B2C iteration.

The Store represents a pivotal shift in OpenAI’s strategy, marking its transition to a product-centric approach. This move is not just about creating a marketplace for AI-driven applications but is also a strategic step towards market domination in the AI application sector. By launching the GPT Store, OpenAI positions itself to control a key distribution platform in the AI ecosystem, showcasing its advanced AI models while simultaneously diversifying its income sources beyond research grants and partnerships.

Impact on Developers and Innovation

The launch of the GPT Store is an opportunity for developers to innovate and experiment with AI-driven applications. However, it’s essential to temper expectations regarding the financial gains for individual creators. The platform serves more as a testing ground for new ideas and applications, offering insights into user preferences and application trends. It’s a chance for creators to contribute to the AI ecosystem, with OpenAI learning from these contributions to enhance its offerings and understand the market better.

The Real Potential of the Store

While the GPT Store opens doors for developers to showcase their creations and potentially earn revenue, its primary role seems to be as a strategic tool for OpenAI to expand its influence in the AI market and learn from the creative endeavors of its user community. The focus is on advancing the AI field and exploring new possibilities rather than on maximizing immediate financial returns for creators.

The Store is a tool by OpenAI for OpenAI. It is not a new gold rush, and it will not create new millionaires. It will though, give millions of ideas to OpenAI.

The Only Opportunity Requires Development

BYOA – Bring Your Own API

The true area of interest within the GPT Store lies in its ability to enable developers and companies to integrate their APIs into custom GPTs. This capability opens the door to specialized services, such as unique data fetching abilities and executing actions on behalf of the user. These integrations could create unique value propositions that transcend traditional app functionalities and allow these services to charge their own fees separate from the GPT Store revenue share.

Zapier, for example, has seen a rise in premium usage thanks to its integrations, from plugins to actions, that have allowed developers to build great multi-service integrations without the need for extensive middleware development.

Some of the coolest GPTs we have seen so far involve external services doing the heavy lifting. For example:

  • Canva: Interacts with its own services to generate graphics, slides, etc. It can’t understand what it has done, but it does generate cool drafts.
  • VideoGPT: By Veed, helps create short videos given a topic, asks you a few questions, and redirects you to their site to continue editing.
  • Consensus: Searches through an external knowledge base of 200M papers. Instructed to help write new papers with accurate citations (though this doesn’t always work).

There are many more examples of what I consider successful GPTs. And they are that because they go beyond the GPT framework to create something new, as a mixture of services. Where GPT-4 is the orchestrator, but the added value comes from the integrated service or knowledge base. In short, the success of a GPT depends really on the quality of the tool that is feeding into the conversation, and not on the prompt engineering of the instructions, or the documents loaded.

Limitations and Creative Solutions

However, these opportunities come with their own set of limitations. like plugins, these integrations have limitations for user authentication and recording of interactions outside of the GPT to reflect back on conversations (as the only way the GPT starts acting is if you message it).

This constraint requires developers to think creatively about how to provide value within these boundaries, challenging them to innovate within the framework of the GPT Store’s capabilities.

Recapping the Barriers of the GPT Store

Revenue Share Limitations

Unlike traditional app stores or creator networks where developers might expect a revenue share of 70–90%, the GPT Store’s model will be markedly different. The likely share for creators is estimated between 10–20%. This lower percentage stems from the fact that it’s OpenAI’s computational resources being leveraged, not the user’s hardware. This distinction fundamentally alters the dynamics compared to other marketplaces.

Market Size Constraints

The requirement for a ChatGPT Plus subscription to access custom GPTs significantly narrows the Total Addressable Market. With ~250,000 Plus subscribers, the reach for creators is limited compared to the 180 million ChatGPT users. This constraint poses a significant challenge for creators aiming for widespread adoption and profitability.

Discovery and Differentiation Challenges

Another critical hurdle is the discovery of custom GPTs within the store. The path to ensuring that a developer’s GPT stands out among potentially thousands is not clearly defined. Moreover, the ease of replicating GPTs means that creating a unique offering is more challenging, diluting the competitive advantage for developers.

A Note on the Horizon

While we’ve covered some of the capabilities OpenAI’s new GPT Store promises, it’s worth mentioning that full details are still behind the veil of missing documentation and lack of usage. As a result, keep an eye on this space — this article may undergo updates to reflect new insights as they come to light.

We invite you to dive into it even deeper. Build, share, and perhaps even monetize your AI on the Store. Particularly if you are a developer, or your company has an API or service that could use the store to create engagement.

As the old adage says… when everyone digs for gold, sell shovels.

Sources & Additional Resources

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓