Egma launches open-source voice agent CI platform

Voice agents are rapidly moving into high-stakes production environments, making manual dial-and-speak QA completely unsustainable and creating an urgent need for dedicated voice CI/CD infrastructure.

• –Automated regression testing: Defining test scenarios in Markdown alongside code bridges modern Git-driven CI/CD workflows with voice application development.
• –Multimodal evaluation challenges: Testing spoken dialog introduces distinct failure modes—including latency spikes, turn-taking, and transcription errors—that traditional text evals miss.
• –Open-source and BYOK posture: Providing an open-source codebase with support for developers' own STT, TTS, and LLM keys eliminates vendor lock-in and mitigates enterprise data privacy concerns.
• –Production-to-test feedback loop: Pairing synthetic pre-merge simulations with live call observability allows teams to seamlessly turn edge-case production failures into regression tests.