The end of 2025 marked the conclusion of my 8-year tenure as Principal Investigator at Backyard Brains. It has been an immense privilege to lead this mission. We set out to make it possible for students to learn neuroscience by building spiking neural networks for robots, and I’m very excited about the final product: SpikerBot (https://t.co/Simv9rrLSH). For an extraordinary example of how to teach neuroscience with SpikerBots in elementary school, see LodeLu (https://t.co/HbilK2vIzo).
But this year, I'm going in a different direction. On January 1, I founded Embrained, LLC. We are developing a software platform that democratizes access to embodied AI. We produce a range of affordable robots capable of collecting large, high-quality datasets, and we design architectures that allow standard PCs to control complex agents.
Embodied AI agents learn to understand your home as maps of traversable nodes. This allows us to play "Graph Games" by simply tweaking the math of the robot's world. If you increase the cost of the living room nodes, suddenly The Floor is Lava. If you drop a high reward on a specific patch of nodes, you create a Conditioned Place Preference. If you mark nodes with few neighbors as preferable, you get wall-hugging behavior (Thigmotaxis). The games you can play with your embodied AI agents are infinite.
I'm betting everything on this platform. I'm looking for early adopters who want to help me test the software and build the first generation of home agents.
Join me at https://t.co/rAfslzYyMO
Dr. Christopher Harris, Founder & CEO, Embrained, LLC
Massively grateful to everyone at the @FieldsInstitute and my dear friend @neurodidact in particular for the opportunity to participate in the Workshop on Multiscale Methods and Emergent Phenomena in Neuroscience this past week. Excellent video recordings of all talks are already available, mine here: https://t.co/4WMvakkckX 1+1=1
To advance foundational AI, we must solve continual learning and catastrophic forgetting. New research by our team introduces Nested Learning (NL), a paradigm that views an ML model as a system of nested optimization problems.
This approach unifies architecture and optimization, creating a deeper computational capacity for learning. This is a crucial step toward creating models with the continual learning abilities seen in the human brain.
More in the blog by Vahab Mirrokni and Ali Behrouz: https://t.co/nVkcCl1Cax
Read the NeurIPS 2025 paper: https://t.co/ArnklW53LI
Friends! After 20 years of work, my neurorobot project, SpikerBot, is finally approaching launch. The official Kickstarter is planned for January, 2026.
We've just launched the official landing page. You can now sign up for a $1 reservation, which guarantees special access and a 40% launch-day discount.
SpikerBot lets you embody almost any brain by simulating it in an app that controls the robot via Wi-Fi - a concept I've been obsessed with for two decades.
I've seen students work happily for hours building and testing brains in the app. As AI systems become ubiquitous, it's time we all have a tool to explore how they converge with, and differ from, our living, loving, human brains.
#SpikerBot #Neuroscience #AI #Robotics #EmbodiedAI