Theres a pretty solid interactive no code builder at https://chat.openai.com/gpts/editor where you can toggle different capabilities (browsing, code interpreter, and dall-e), and upload 20 files for a hands off rag implementation. To make the GPT itself theres an interactive window where you can chat with an “agent” on one half of the screen telling it what you want the GPT to do, how you want it to respond, etc. and on the other side of the screen it interactively updates the GPT in real time where you can test it and it’s writing a detailed system prompt and everything on the backend (which you can edit in an advanced editor tab). It’s a pretty solid UX and fairly intuitive and easy to use for basic assistants but theres also an advanced editor that lets you customize prompts and add actions and hook it up to third-party APIs.
Overall, A lot of the prompting is abstracted away behind the scenes but they’re not really full featured web apps, they’re more like specialized chatbots. The benefit for creators that I see is opposed to having to deal with hosting a chatbot themselves and paying for API costs, they get with access to OpenAI covering the usage cost and giving them features that are hard to implement like code interpreter and features they would build themself like a knowledge base for RAG + access to external APIs all with a rev-share for the creator + data for OpenAI.
