What should an AI-native API workspace do better than today’s API clients?

API Circle Studio is launching tomorrow.

I’m building it around one idea: API workspaces should work more like code.

That means:

• Collections and environments should be readable and git-diffable

• API changes should be reviewable through pull requests

• Local mocks should be part of the workflow, not a separate setup

• AI clients should be able to read, author, and run API workflows safely through MCP

I’m curious to hear from API developers and platform teams:

1. What frustrates you most about current API clients?

2. And what would make an AI-native API workspace genuinely useful instead of just another request sender?