AI providers, models, prompts, rate limits and live connection status are now managed in the application console, which reads and writes the configuration the platform actually uses.