EchoForge launches on-device voice agent SDK and takes $25M seed

AI · 4 min read

EchoForge's new SDK packages a modular on-device voice stack that includes wake-word detection, streaming ASR, compressed intent models, and a micro-LLM optimized for short, goal-directed dialogues. The company emphasizes privacy, with models running entirely locally unless explicitly configured to use cloud services.

Investors in EchoForge's $25 million seed include EdgeFund and multiple telecommunications strategic partners. The funds will be used to expand platform support for Android, iOS, and proprietary Linux builds, as well as to build enterprise tools for voice model fine-tuning and evaluation.

EchoForge is targeting consumer-device makers and automotive partners who need responsive, private voice agents. The startup also announced partnerships with two mid-tier smartphone OEMs to embed the SDK into upcoming mid-range devices.