Egra

Egra is a neurotechnology and artificial intelligence company developing foundation models for electroencephalography (EEG) and other biosignal modalities. Its stated aim is to align AI systems to human neural responses by converting involuntary biosignals into reward signals usable in model training, an approach the company calls Reinforcement Learning from Biofeedback (RLBF). The company frames this as an alternative to conventional human-feedback signals such as clicks, thumbs ratings, star ratings and written rationales, arguing that neural responses are denser, personalizable and can surface subconscious reactions such as the uncanny valley effect.

The product roadmap has three components: a foundation model that maps multimodal biosignals to error and preference signals across text, audio and video; a capture stack of lab-grade and consumer-grade hardware that records biosignals time-locked to stimulus onset with millisecond-accurate alignment; and an alignment platform that converts raw biosignal data into a reward stream intended to plug into the post-training pipelines of frontier AI labs. The company is reported to have around six employees and is recruiting AI researcher, engineer and intern roles in New York City.

Business model

Egra develops proprietary models, biosignal capture hardware and a pipeline that delivers reward signals into the training workflows of AI model developers.

Traction

Reported total funding of $5.5 million and a team of approximately six employees.

Latest developments

In April 2026 Egra was reported to have secured $5.5 million in new investment capital, an announcement also carried on the company's own website. The proceeds were described as funding further development of its EEG foundation models and expansion of the team.