Applied Brain Research (ABR) develops AI for edge devices using state space models – a technology they created and patented, which the company frames as “the post-Transformer era”: real-time speech-to-text transcription, natural expressive text-to-speech generation with no delay, speech-to-intent understanding, language translation, and biosensing applications, all running entirely on-device without cloud dependency.
On-device, always-real-time processing is the genuine technical differentiator – most speech AI depends on cloud round-trips that introduce latency and require network connectivity, while ABR’s state space models are specifically engineered for faster processing with reduced complexity on time-dependent signals, delivering ultra-low power consumption that extends battery life on wearables and AR devices where power budget is a hard constraint, not a nice-to-have. A privacy-focused architecture where audio never leaves the device is a direct consequence of that on-device design, not a bolted-on privacy feature.
ABR targets AR glasses, wearables, smart home devices, robotics, and companion robots as deployment targets, demonstrating first silicon hardware acceleration for their state space model technology in November 2025 – a concrete engineering milestone, not just a software roadmap claim. This is technology licensed to hardware and product companies rather than a self-serve consumer product; pricing isn’t disclosed.
