DreamServer voice agent framework goes fully local

Fully local, multi-agent voice systems have been the missing piece for developers who want voice AI without handing data to cloud providers — DreamServer makes it a one-command install.

• –Eight specialist sub-agents with priority-weighted intent detection means calls route correctly without hardcoded if-else logic
• –Shared `CallData` state persists across agent handoffs — a common failure point in naive multi-agent voice designs
• –TTS output filtering strips internal markup before it hits the speaker, solving a real usability gap in local TTS pipelines
• –n8n webhook integration turns completed calls into structured tickets automatically — practical for real deployments
• –Competes with Pipecat and LiveKit Agents but differentiates as a batteries-included platform rather than a composable framework