Or replacing "For example, If you're charging for API requests" to "For example, If you're charging for LLama AI Model API requests".
Heck, I had to review a doc change at work that was pretty stupid. Like one thing we offer is an S3-compatible endpoint. But someone thought we should clarify that you can upload AI models there too and all our docs should include an "AI developer" section for how to upload a blob that also happen to be a model or a lora or whatever.
Because apparently Minio is for AI these days: https://min.io/
When the AI craze stated, so many people in my company came to me asking “if we can run AI workload”? another thing we offer is fairly generic compute meant for your average web applications or micro service etc. Initially I said “I don’t think so. We don’t have GPUs nor do we have any ability to express hardware requirements beyond CPU and Memory. We’ll need to do some work to include GPU into that”.
Then hilariously I learned that you don’t need GPUs or ASICs to be able to run “AI workloads”. If your compute allows you to call OpenAI rest APIs, then you’re also “AI Ready”.