Flybrain presents a novel, technically grounded experiment: a live simulation of the Drosophila melanogaster male CNS connectome (165,122 neurons, 10.2M connections) deployed on-chain to autonomously interact with web interfaces and launch its own token, $FLYBRAIN, on Robinhood Chain (EIP-155:4663). The website (web_1) explicitly documents its fidelity to the publicly released HHMI Janelia/Cambridge/Google dataset, its biophysical neuron model (leaky integrate-and-fire), and its real-time sensory-motor pipeline — mapping browser screenshots to retinotopic input and cursor output via identified descending neurons. It transparently discloses limitations: no language capability, no goal-directed behavior, reliance on a narrator LLM for social posts, and an enforced allowlist of sites (Wikipedia, arXiv, etc.) due to technical incompatibility with major platforms.
The project’s core thesis — that a biologically accurate neural circuit can meaningfully interface with and act upon the internet — is compelling but lacks verifiable evidence of functional economic agency. While the token contract is confirmed on-chain (api_1, api_2), the website states the fly “reliably gets the description, often the ticker, rarely the name” and that a script completes missing fields (web_1). There is no evidence of autonomous, reliable, or economically consequential action — e.g., sustained interaction with DeFi protocols, collateral management, or redemption mechanisms. No documentation, explorer data, or API records describe token utility, holder rights, treasury, or governance (api_1 notes only meme tags and supply). Critically, the rubric’s primary specialist criteria — collateral quality, redemption, liquidation, or yield — are not applicable, as Flybrain is not a stablecoin, RWA token, or yield-bearing instrument. Its value proposition is experimental and narrative, not economic infrastructure. This makes the primary investment thesis non-verifiable with the supplied evidence, triggering the hard cap limiting scores to ≤6 and mandating low confidence.
Overall score: 5/10 Confidence: Low
Orbio.so positions itself as a discount AI inference relay, enabling users to access models like Claude and GPT-6 via a single OpenRouter-compatible key while claiming discounts sourced from token holders and API key suppliers. Its website (web_1) details real-time deposit mechanics, latency metrics, and usage-based pricing — but provides no verifiable description of how discounts are generated, enforced, or redeemed. The token $ORBIO is reported across two API sources (api_1, api_2) as having a total supply of 950M with zero circulating supply (api_1) or 950M circulating (api_2), a direct contradiction that remains unresolved. Neither source explains custody, redemption rights, or economic claims about fee conversion into credits (api_2 states “Half of trading fees are converted into inference credits and paid to holders, in dollars”), yet no on-chain evidence, treasury address, or mechanism for credit issuance is supplied.
Why is there no verified collateral or redemption path? Because the project’s core value proposition — discounted inference credits backed by token holdings — lacks any disclosed smart contract logic, oracle feed, or custodial arrangement. The contract address on Robinhood Chain (0xaa07…28a3) is listed but unverified; no audit, bytecode analysis, or functional test is present in the evidence.
Do holders actually receive credits? No evidence confirms distribution mechanics, timing, eligibility, or claimability. The claim appears aspirational, not operational.
The absence of documentation, repository links, or technical specifications further limits assessment of operational security or code permissions. Confidence is low due to contradictory supply data, missing primary-thesis evidence, and reliance on unverified API projections.
Overall score: 5/10 Confidence: Low