I am working on using OpenAI as part of a product (grid.is) and have a few requests for better key management and operational safety while using the API:
I would like to be able to generate API keys for organizations, not just users. If the user with the keys that are used in production leaves the OpenAI organization we must switch over to new keys issued by a new user, this causes risk to the availability of the features of the product that rely on OpenAI. (I tried to solve this with a shared user for our organization, but it requires a phone number that has not been used before, and not a VoIP number, so I ended up abandoning that approach).
Secondly, I would like to be able to add a description to the keys when they are generated. We use a handful of keys (prod, staging, local dev) and it’s important that it’s clear which key is used in which environment when we’re managing them in the OpenAI web console.
Thirdly I would like us to be able to have a second organization for staging and local dev, or some sort of separation from the organization that we use for prod. I view this both from usage perspective (an error in local dev should not impact the availability for production), as well as from cost perspective (we are willing to allocate a higher limit for production than for staging or dev). One way to approach this would be to allow the linking of organizations or having sub-organizations.
I’m looking forward to improvements in this area that improve our management of the OpenAI integration!