FEAGI
Open-source engine for adaptive robot intelligence
FEAGI 2.1.43 is now available.
The bigger picture
It took decades for CNNs, RNNs, and Transformers to emerge and fuel the AI revolution we live in today.
Each breakthrough required deep expertise, institutional resources, and years of focused research. The people who could push AI forward were always a small, credentialed few.
Imagine a tool that lets anyone — regardless of background — design, experiment with, and iterate on AI architectures. Not prompt engineering. Not fine-tuning someone else's model. Actual architecture design.
FEAGI makes that possible. Its modular, intuitive, and biologically-grounded architecture turns the hardest part of AI research into something anyone can explore, build on, and share.
A platform built to grow with its community
FEAGI is more than a neural engine. It is an open ecosystem that compounds in value as more researchers, engineers, and builders contribute brains, agents, and integrations. The network effect is intentional and architectural.
Ecosystem effect
A platform where anyone can build, share, and benefit
FEAGI enables anyone, from domain experts without a programming background to seasoned engineers, to design neuromorphic AI solutions for robots and beyond. Under the Apache 2.0 license, every brain architecture, agent, and controller built on FEAGI can be shared freely and built upon by the community. The more people contribute, the more powerful the platform becomes for everyone.
Academia meets industry
Peer-reviewed science. Production-ready engineering.
FEAGI is grounded in multiple published IEEE papers and actively adopted in university research programs. At the same time, its deterministic, auditable execution model and cross-platform deployment targets make it ready for commercial robotics, edge AI products, and industrial automation. It is one of the few platforms that can credibly serve both a PhD thesis and a product roadmap.
Designed for difficult environments
From embedded microcontrollers to cloud-scale deployments
Most AI platforms assume a GPU server. FEAGI assumes the opposite: intelligence must run where the robot lives. Its event-based spiking architecture is inherently energy-efficient, with no continuous matrix multiplications and no idle compute. The same brain runs on an ESP32 microcontroller, a desktop workstation, a Docker container, or a Kubernetes cluster, without rewriting a single line.
Modular by design
Composable I/O. Extensible agents. Embodiment agnostic.
FEAGI's peripheral nervous system (PNS) model provides standardized sensory inputs and motor outputs that any embodiment can plug into. Add a new robot, simulator, or sensor without touching the brain. New controllers follow a published standard and can be contributed back to the open-source library or distributed through the marketplace.
Escape the limitations of LLMs and derivative AI models
What can FEAGI do that others can't? The sky's the limit. Because of its unique brain-based structure and varied input sources, it can learn to read printed text, differentiate between objects, and even control robots. The possibilities are endless; the development is up to you.
The ecosystem flywheel
FEAGI anchors a four-layer ecosystem. Each layer grows the next: more brains in the Hub make NRS more valuable; more NRS users contribute more controllers; more controllers expand the addressable hardware market.
OPEN SOURCE ENGINE
FEAGI
The neural execution engine at the foundation. Apache 2.0, free for academic and commercial use. Every brain, agent, and controller in the ecosystem runs on FEAGI.
DESKTOP APPLICATION
Neurorobotics Studio
The visual workspace for FEAGI. Design neural circuits, connect to simulators or physical robots, train with live sensor data, and visualize brain activity, all in one desktop application.
MARKETPLACE
Brain Hub
A growing library of community-built and curated brain architectures. Download a ready-made brain for your robot, share your own, or publish controllers through the marketplace.
INTEGRATION LAYER
Embodiment Controllers
Open-source controllers connecting FEAGI to 20+ physical robots, simulators, BCI devices, and IoT hardware. Follow a published standard to add your own, and optionally publish it to the marketplace.
From brain design to live deployment: FEAGI in action
These tools are part of the ecosystem FEAGI enables, visualizing neural activity in real time, wiring cortical circuits, and training adaptive AI models without writing low-level code.
3D Brain Monitor
Visualize the entire neural network in an interactive 3D environment, monitor brain activity in real time, and stimulate specific brain regions on demand.
Circuit Builder
Design and rewire brain regions to shape behaviors, gain deep insights into artificial brain functionality, and analyze intricate wiring diagrams.
FEAGI Trainer
Leverage supervised and unsupervised learning to train the brain, providing labeled data or letting it autonomously learn as fitness is assessed.
Seamless integration with leading simulation & robotics frameworks
FEAGI is designed to work effortlessly with industry-standard simulation and robotics platforms, enabling rapid prototyping, testing, and deployment of intelligent robotic systems.
GazeboSim integrates seamlessly with FEAGI, bringing neuromorphic intelligence to physics-accurate robotic simulations with real-time control.
Learn and share with others
Fork the GitHub repo, build something fresh, and share it with the world.
Chat with other developers, get support, and collaborate on projects.
Follow us for updates, networking, and professional growth opportunities.
Easily add your project to the Hub and share it with users.
An ecosystem built to scale
FEAGI creates a compounding network effect: every robot controller, brain architecture, and simulator integration added by the community increases the platform's value for every other user. Open source accelerates adoption. The marketplace closes the loop on monetization.
Apache 2.0
Permissive open-source license, free for academic and commercial use
Multiple IEEE papers
Peer-reviewed publications across transactions, circuits, and computing conferences
3 SDKs
Python, Rust, and Java that integrate into any stack
20+ embodiments
Physical robots, simulators, BCI devices, IoT, and microcontrollers
For developers and researchers
Explore the full technical reference on BrainsForRobots.com: SDKs, Rust core performance, embodiment controllers, deployment targets, and peer-reviewed academic citations.
Full FEAGI referenceFor investors & industry partners
FEAGI is the foundation of Neuraville's commercial platform.
The open-source engine drives developer adoption and research validation. Enterprise licensing, OEM integration, and marketplace revenue are built on top. Learn about the full investment thesis, traction, and go-to-market on the company page.
Company overviewgithub.com/feagi