This post has been deleted - 101

GPT builders
1

As a developer spending much time building custom domain-specific GPTs (non-API) on the GPT store and promoting these GPT through various media channels. I would like to ask OpenAI about the following questions regarding non-API custom GPTs:

  • Is there a difference in terms of how OpenAI sees developers using their APIs vs developers only using the web custom GPT tools?
  • Will there be any way for custom GPT developers to “own” their custom GPTs (such as instructions) in a way that an GPT API developer has ownership over their own GPT wrapper?
  • Will there be options for custom GPT developers to monetize or to access user information (e.g. getting access to emails of users that used my custom GPT)?
  • Should developers serious about building apps on the OpenAI ecosystem shift towards using APIs instead of relying solely on custom GPT (e.g. if I want to do a company dedicated to custom apps on OpenAI)?
2

I’m going to answer one of the questions, specifically how OpenAI possibly views developers. So I’m going to tell you that for example in other development environments such as video game engines, there is something called Blueprints, a way to code a game without having to deal with programming languages like c++, this talking about UE5 (unreal engine 5), so what blueprints does is to bring people who do not know how to program to create their games, as you can see the goal is the same, whether you use c++ or Blueprints.

So in this context the same thing happens, the Builider gpt helps in productivity, you will have a functional gpt just saying in natural language, “I want you to be an expert in German cooking”, and you have something created. Now, there are disadvantages, and that is that in order to compete with other gpts that specialize in the same thing, you will have to give your gtp some particularity that makes it better. In such a case the builider gpt is a tool that does not allow you the flexibility of using the API directly from OpenAI. So we could say that it is just a tool, that if you are an expert you can get the best out of it. For example, the key is to give precise instructions, “Expert in German cooking”; but you will have to tell it that it will not give recipes from other countries, if you don’t give it the instruction it will talk about anything other than German cooking. You tell him what to do and what not to do.

3

Your questions reflect a deep interest in the development within the OpenAI ecosystem. Let’s take a closer look:

  1. API Usage vs. Web Custom GPT Tools: OpenAI likely views the use of the API and the Web Custom GPT Tools differently. The API allows for more flexible and deeper integration of GPT into one’s own applications and systems, while the Web Custom GPT Tools are probably intended for simpler, direct applications. It strongly depends on your project and its requirements which option is better suited.

  2. “Ownership” of Custom GPTs: Regarding the “ownership” of models created via the GPT tools, it’s unlikely that OpenAI offers the same kind of ownership rights as with independent API development. The GPT tools might be seen more as a platform on which you create your own models, but they are based on OpenAI’s infrastructure.

  3. Monetization and User Data: As for monetization and access to user data (such as email addresses), there are currently no clear guidelines. This heavily depends on privacy policies and terms of use.

These are areas where the landscape is constantly evolving, so it’s advisable to stay updated with OpenAI’s latest offerings and policies.

1 Like