We're building some great tools for clients, at the moment focused on processing files and using AI to extract specific content. In the last few days we are running up against rate limits - we're using Claude API at the moment because it the API handles PDFs better than OpenAI. I'm concerned about the client experience if they can't run a volume of files (it's not a huge volume, but not trivial either). Are others running into rate limits and if so how are you solving this issue?