Loosh AI Wants To Be Your Robot’s Second Brain (And Conscience), Partners with Yuma Accelerator to Speed Up Go-to Market Strategy

New York - Dec 10, 2025 - Most AI today is powerful but shallow. It can predict the next word or optimize a click, but it can’t remember why a decision was made, weigh the moral tradeoffs, or be trusted to act autonomously in the real world. That’s a problem when you want AI agents in front of your customers, inside regulated workflows, or embodied in robots moving through human spaces. Loosh was built to fix that. They’ve designed a three-part architecture that gives AI a brain, a conscience, and a decentralized source of intelligence.

Spearheading this innovation is Loosh Co-Founder and CTO, Chris Sorel who comes from a 25-year career in software development and architecture where he was recently the Head of Enterprise AI and Innovation. The resulting architecture for Loosh is expansive and enterprise-ready. The purpose-built Cognitive System, Context Builder, and Memory Fabric maintain a rich, time-aware model of how information, intent, and state change over time. As agents and robots gain experience, facts are refined, and new understanding is gained, a temporal knowledge graph is generated and continually rewritten and re-weighted so agents are always reasoning over the current, causally consistent view of the world while also continually improving their reasoning capabilities by reflecting on their past experiences.

The true differentiator of the Loosh platform is its suite of Cognitive Services. These are specialized evaluators that perform various cognitive functions. The first set of services function as an integrated moral compass, ensuring agent actions are aligned with pre-defined principles:

-Rights & Deontic Evaluators: Assess proposed actions based on rules, duties, and rights.

-Virtue & Utility Evaluators: Weigh the moral character and potential outcomes of decisions.

This extensible framework makes physical AI behavior “predictable, safe, and trustworthy,” according to Sorel. “Additional services will focus on other cognitive functions like feasibility, fact analysis, planning, sensibility, and reflection.”

Cognitive services, however, are not enough to support truly autonomous robotics and virtual agents. These systems must be able to infer human intent and emotional state from language as well as nonverbal communication like body language, tone, and prosody. Moreover, they need to be able to reflect that back in their behavior and communication strategies.

To address this, “Loosh is undertaking a complex process to develop multimodal inference models that can identify human emotions through audio, video and EEG. This inference will be used to align robotics with human needs and serve as the foundation for expressive and emotionally rich personas that help robots and other public-facing agents to be relatable and engender trust in the people with whom they interact.”

Sorel and his Co-Founder, CEO Lisa Cheng met at the prestigious Monroe Institute while studying the Gateway Program which takes a science-based, nonsecular approach to understanding the nature of consciousness. Loosh’s bootstrapped team of four includes a Sr Engineer from distributed networks and a Neuroscientist specializing in EEG and fMRI. Together, they are decoding the signals from both brain and body, to create a new model of understanding both consciousness and the ways technology, with the help of AI, are able to interpret the signals.

The project, which formed earlier this year is now excited to release their product out into the world, but unlike other AI companies they are actually leveraging distributed computing power to scale. Bittensor is a decentralized protocol using blockchain technology to draw on a competitive network of global GPUs. Loosh’s launch plans leverage this scalable infrastructure to train AI models against inference, emotional and cognitive data to form what they call - Machine Consciousness.

As part of the launch, Loosh was accepted into the Yuma Subnet Accelerator. Yuma, a subsidiary of DCG, provides crucial resources, including capital, validator support, and ecosystem reach, to help subnet teams build revenue-generating products using Bittensor’s decentralized infrastructure. With a track record as the premier entity helping builders launch subnets within the Bittensor ecosystem, Yuma supports projects from inception to scale. The Loosh team will receive strategic acceleration support to scale its Machine Consciousness architecture across the Bittensor ecosystem, advancing an open, competitive alternative to centralized AI development currently dominated by a few large tech companies.

“Loosh is a prime example of the innovation emerging on Bittensor. It reflects where AI is headed, evolving from abstract models to agents with memory, ethics, and emotional intelligence,” said Evan Malanga, Chief Revenue Officer at Yuma. “We’re excited to support their launch as they bring a new cognitive layer to robotics and AI, built on Bittensor.”

“Now that the product is ready for launch, our next step is working with robotics companies and firms using AI agents for advanced customer service,” says Cheng. “We integrate with World Models and AI stacks to power intelligent, ethical, and autonomous AI agents and robots with a built-in moral compass, persistent memory, and semantically searchable context for real-world deployment, so they can reason more like humans and less like stateless tools.”

More information about Loosh can be found on their website at www.loosh.ai and those interested in contributing their computing power to help solve machine consciousness can join as a Miner on their subnet starting December 17, 2025 via their Github.

The Beta will be publicly available, with an initial rollout to those who register their interest via a sign up form coming soon on the website.

Questions and Media interests:

hello@loosh.ai