Feature request: Direct tool execution with a short result message
Hello OpenAI team.
I noticed that ChatGPT works very smoothly with image generation. When I ask it to create an image, it directly starts generating the image and then shows the result.
I think this same behavior would be useful for other connected AI services and tools too.
Idea:
ChatGPT could have a direct action mode for connected tools and AI services. For example, if a user asks:
“Generate a voice with ElevenLabs”
“Create a short video”
“Make music”
“Edit this file”
“Use this website or tool to complete the task”
ChatGPT should directly call the needed tool or integration when the request is clear. After the tool finishes, ChatGPT can reply with a short result message, for example:
“Done. Here is the generated audio.”
“Done. Here is the edited file.”
“Done. Here is the video.”
The important part is that ChatGPT should not only say “Sure, I will do that” without actually starting the action. It should perform the action first when possible, then give the result.
Why it would help:
This would make ChatGPT feel more like a real assistant. The user gives a clear command, ChatGPT performs it, and then returns the result.
How it could work:
-
Users could connect approved third-party AI services.
-
ChatGPT could detect when a task needs a specific tool.
-
If the user’s intent is clear, ChatGPT runs the tool directly.
-
If permission, payment, login, or safety confirmation is needed, ChatGPT asks first.
-
There could be a setting like “Direct action mode” to reduce unnecessary confirmation messages.
This would make ChatGPT faster and more useful for creative workflows.