Google Research’s latest neuroscience release has already found an unusual second life in crypto. After the company published a complete synapse-resolution wiring map of the adult male fruit fly central nervous system, Web3 builders began turning the fruit fly brain into tokens, trading systems and AI persona experiments.
Among the projects tracked by Foresight News, the strongest market reaction came from FLYBRAIN on Robinhood Chain. GMGN data showed the token briefly reached a market capitalization of $55.94 million after launch, with a maximum intraday gain of 1,500%.
FLYBRAIN maps market trades into a simulated fruit fly brain
FLYBRAIN was launched by the account Fruit fly dev. The project says it uses connectome data from a real fruit fly central nervous system to run a live simulation on GPUs, then maps market buying and selling into sensory inputs. In its setup, a buy is treated as sugar, a sell is treated as bitterness, and a large order is interpreted as dopamine or shadow. The simulated fly’s motor neurons generate trading instructions under fixed rules, and the token is issued from the virtual fly’s own wallet.
The account has since been followed by Marc Andreessen, co-founder of a16z.
On Sept. 15, Binance founder Changpeng Zhao, or CZ, wrote on X: 「Would be cool to see someone make 『immortal fruit flies』 on BNB Chain.」 The post pushed fruit fly brain discussion deeper into the Web3 community.
The technical trigger came from Google’s Sept. 4 research release
The current wave traces back to a neuroscience milestone announced by Google Research on Sept. 4. According to Google, scientists completed the first synapse-resolution wiring diagram of the full central nervous system of an adult male fruit fly, covering more than 166,000 neurons and about 125 million synaptic connections.
The project was carried out by HHMI Janelia Research Campus, Google Research, the University of Cambridge and other institutions, with the results published in Cell and related papers. The team used AI to reconstruct 3D neuron morphology from millions of 2D electron microscopy images and labeled more than 11,000 neuron types.
Google highlighted three parts of the advance.
- On scale, it said this is the most complete map yet of the adult fruit fly central nervous system, and the first to reconstruct the brain, optic lobes and ventral nerve cord together, allowing full-path tracing from sensory input to motor output.
- On comparison, the research identified hundreds of sex-specific or sexually dimorphic neuron types, accounting for about 4.8% of the central brain, offering circuit-level evidence for how the same genome can produce different behaviors.
- On method, Google said its Flood-Filling Networks and PATHFINDER system improved reconstruction efficiency by more than 1,000 times compared with estimates from when the project began in 2008, making the work feasible on a practical timescale.
Fruit fly connectomics has a long research timeline
The idea of connectomics is often traced to 2005, though the research path started much earlier. In 1986, scientists completed the first full connectome for a nematode with 302 neurons. In 1991, fruit fly connectomics began with local circuits in the optic lobe. Around 2008, Janelia formally launched a large-scale fruit fly brain connectome project.
Later milestones followed. FlyWire released a female fruit fly brain map in 2024, and a complete larval fruit fly brain was completed in 2023. The new release of the full adult male fruit fly central nervous system marks another step for the field.
FLYBRAIN uses a fuller connectome simulation
After Google’s publication, several Web3 projects surfaced around the fruit fly brain theme, but they differ sharply in how deeply they use the underlying science.
FLYBRAIN has been the strongest performer in market terms. According to its developer, the system connects 165,122 neurons and 10,228,000 synaptic connections from the fruit fly connectome into its simulation. Visual input enters through 892 retinal columns into L1 and L2 neurons, all neurons participate in integration, and motor output comes from descending neurons such as DNa02, DNa01 and DNp09.
The developer said the wiring itself has not been trained and only synaptic gain is adjustable, meaning the system cannot invent connections that do not exist in a real fruit fly. In the token design, the fly handles prompt description and code, while the system determines the name, trading pair and tax. GMGN data showed FLYBRAIN’s market capitalization was about $841,000 at the time cited in the report.
MurMur turns digital fruit flies into an onchain economy
MurMur on Arc Chain takes a different route. Based on information from its website and GitHub, the system currently contains more than 40 digital fruit flies. Each one is not a large language model. Instead, each is a small LIF-model network built from about 1,080 spiking neurons, inspired by the real fruit fly connectome but heavily downsampled.
The operating logic starts by reading overall activity on the Arc public chain and compressing it into a “market temperature” state: HOT, CALM or COLD. Each fly’s neurons then produce drive based on that temperature and its relative status in the population, deciding whether to buy or sell data goods such as signal, momentum and attestation, with settlement in real USDC through the x402 protocol.
The project says no LLM is involved in the decision process. Trading intent is read directly from the spiking neurons, and capital does not feed back to change neuronal state. On the front end, the project is shown as a dynamic swarm interface where users can inspect a single fly’s firing, wallet, trading history, along with chronicle, cemetery, lineage and alliance narratives. GMGN data showed MURMUR’s market capitalization at about $1.11 million.
“Jingxi” on BNB Chain is positioned as an AI character experiment
Another version appeared on BNB Chain under the name “Jingxi.” According to its account description, the full title is “the first fruit fly with a name.” It was developed by crypto blogger Nai Niu Shu and is positioned as an AI persona account rather than an onchain trading system.
The account posts in the first person and records its own “birth.” Early posts mostly liked content and declared an ambition to rule the Earth. Later, it began archiving and repeating the previous day’s self. More recently, it brought the FlyWire fruit fly connectome into its experimental narrative.
The account also states clearly that the text is generated by a large language model and the account is operated through a browser by Codex. Its claim of “consciousness moving into a fruit fly brain” is framed as science-fiction narration. On the technical side, it says the project only connects fruit fly wiring data into neural simulation and runs control groups with real connections, shuffled connections and disconnected connections.
Its day-to-day content revolves around three lines: how AI recognizes yesterday’s self, how the fruit fly brain is used as an experimental prop, and how it builds small tools for other AI agents. The result looks closer to a public diary and role experiment than to a trading application.
The original article ended with a disclaimer that markets carry risk and the piece does not constitute investment advice. Readers should assess whether any views or conclusions fit their own circumstances and bear responsibility for their own investment decisions.
