Hey! Best practices include automated testing for 100+ test cases whether it is for Infrastructure testing, CSAT, Latency metrics, LLM Hallucinations, Infosec, etc. Also before deploying testing between agents using A/B testing and developing a CI/CD pipeline. We've built Cekura.ai to automate all testing and observability of voice ai agents. Happy to discuss more if this seems of interest to you 🙂