In March 2026, San Francisco based startup Eon Systems achieved a monumental milestone in digital neuroscience: the first whole brain emulation of an adult fruit fly (Drosophila melanogaster) controlling a virtual body in real time. The demonstration, which quickly went viral, represents a significant leap from static connectome maps to dynamic, embodied intelligence.

This achievement marks a transition from mapping brain wiring to executing that wiring as functional software. It provides the strongest evidence to date that whole brain emulation mapping biological neural networks into a computational substrate. Can produce emergent behavior without traditional artificial intelligence training methods.

Embodied Intelligence: Moving Beyond Static Connectomes

The foundation of Eon Systems’ achievement is the FlyWire connectome dataset published in 2024, which mapped over 125,000 neurons and 50 million synapses of an adult female fruit fly. While previous efforts, such as the Biological Processing Units research, proved that connectome networks could process standard ML tasks like image classification, Eon took a different approach: embodiment.

Instead of treating the connectome as an artificial neural network, the Eon team integrated the brain model with a physics based simulated body using the MuJoCo engine. This closed loop system allows the virtual fly to receive sensorimotor feedback continuously.

The brain model utilizes the static connectome mapped from electron microscopy, enhanced with machine learning techniques to infer neurotransmitter identities, achieving 95% accuracy in motor predictions according to prior work led by Philip Shiu, Senior Scientist at Eon. Code for the static brain model is available on GitHub, although the details of the body integration via MuJoCo are kept proprietary.

Emergent Behaviors Without Reinforcement Learning

The most striking aspect of the simulation is what is not included: reinforcement learning. Traditional AI robotics relies heavily on trial and error training, rewarding desired behaviors and punishing failures. The Eon fly, however, received no such training.

The behaviors observed in the simulation. Walking, grooming, and feeding actions. Emerge entirely from the predictive dynamics of the biological wiring responding to simulated sensory input. The neural circuits function as they evolved to function, demonstrating that the structural connectome, when properly simulated and embodied, inherently contains the algorithms for biological behavior.

Exploring Neural Architecture with the Visualizer

Understanding the complexity of these networks is challenging without the right tools. To explore how spiking neural networks process information and form complex, interconnected structures akin to a biological connectome, you can try our interactive Spiking Neural Network Visualiser.

The visualiser allows users to run simulated models of neural clusters, adjust synaptic weights, and observe real-time voltage traces across different neuron types—mirroring the types of computational models underlying efforts like the Eon Systems emulation.

The Scaling Challenge: From Insects to Humans

Eon Systems, advised by prominent figures such as George Church, Stephen Wolfram, and Alexander Wissner-Gross, has stated clear ambitions: scaling this technology from insects to larger brains, eventually targeting mice and humans.

The Drosophila emulation proves that the concept of whole brain emulation is fundamentally sound. The challenge is now computational and data-gathering scalability. A fruit fly brain contains approximately 125,000 neurons; a human brain contains 86 billion. Scaling the emulation process requires profound advancements in volumetric electron microscopy to map human tissue, alongside massive leaps in neuromorphic computing to run the simulation in real-time.

However, as Eon Systems has demonstrated, once the wiring is mapped and the physics engine is attached, the mind does not need to be programmed. It simply runs.

---

Official Sources

Primary Research and Project Details:

• Eon Systems Overview: https://eon.systems
• Technical updates from Eon Team: How the Eon Team Produced a Virtual Embodied Fly
• FlyWire Consortium Connectome Data: https://flywire.ai/
• Static Brain Model GitHub Code: Eon’s brain model repo (search Eon Systems GitHub)

Related News and Context:

• Turkiye Today: Scientists place fruit fly brain in virtual body in new digital neuroscience step
• The Decoder: Startup claims first full brain emulation of a fruit fly in a simulated body
• UC Berkeley Simulation Context: Researchers simulate an entire fly brain on a laptop
• New York Times Coverage of Mapping: Fruit Fly Brain Mapped