Veliora: Agentic Biomedical Research Engine
Veliora is a multi-agent biomedical research application, built on Arc and powered by Circle, designed to make disease-focused repurposing analysis more structured, traceable, and economically viable. Today, early-stage biomedical research is often fragmented across literature review, target analysis, pathway interpretation, and manual candidate comparison. This process is slow, expensive, and difficult to audit—especially when teams need to determine whether a signal is genuinely promising or still too weak to advance. Veliora transforms this workflow into a funded, staged research system. A user submits a disease-focused question, funds the job in USDC, and specialized agents perform literature review, candidate screening, pathway analysis, evidence scoring, critical challenge, and report synthesis. Instead of forcing every run into a potentially misleading final answer, the system can return reportable candidates, early-stage hypotheses, pipeline-reviewed signals, or a rejected outcome, depending on evidence quality. What makes the project distinctive is the combination of multi-agent orchestration and a usage-based economic model. Payments are tied to actual task progression rather than flat subscription access, and the final output is a structured research brief with traceable provenance. Veliora is designed for researchers, evaluators, and scientific teams who need a more transparent and reliable way to prioritize evidence-backed signals while preserving reviewability and decision quality.
Category tags:"Veliora is an outstanding submission that stands out for its combination of technical rigor, domain differentiation, and thorough proof-of-work. Circle + Arc integration is deeply core: x402 is used for per-agent-action micropayments at $0.002 per action, Circle batches settlements, and Arc provides USDC-native low-cost finality. The economics slide clearly articulates why this model only works on Arc ($0.002/action is 5x below viable threshold for traditional onchain settlement). Circle/Arc Integration (Criterion 1): Sub-cent pricing at $0.002/agent action is verified. The submission shows 50 paid actions across 5 specialized agents (literature/search, drugdb/fetch, pathway/analyze, red-team/critics, review-service/review) — meeting the high-frequency threshold. ERC-8183 escrow is used to lock USDC budgets before the agent pipeline runs, with auto-settlement after task completion. Two buyer wallets each ran 5 research runs, locking and spending 25 USDC total. Multiple seller wallets received +0.042 USDC each as post-batch settlement evidence. Agentic Model (Criterion 2): The flow is genuinely agentic and multi-agent — not a human paying an API, but orchestrated agents paying other specialized agents per task. Each pipeline stage (literature, drugdb, pathway, red-team, review) is a paid service node. This is a textbook agentic economy use case. Working Product & Demo (Criterion 3): Demo is hosted live at veliora.vercel.app. The video (4:39) shows the complete workflow: user submits research topic (Idiopathic Pulmonary Fibrosis), funds 3 USDC budget, agents execute, and results are returned with traceable provenance. Transaction table in submission shows 27+ rows with individual TX hashes, verify=true, settle=true. Seller post-batch settlement table shows observed balance changes per wallet. This is one of the most comprehensive proof-of-work presentations in this hackathon. Technical Depth (Criterion 4): Multi-agent orchestration, ERC-8183 job escrow, x402 nanopayments per stage, USDC on Arc, staged research pipeline with reject/refund logic. Thoughtful system design with clear separation between buyer wallets, seller wallets, and the orchestration layer. Differentiation (Criterion 5): The biomedical research vertical is highly original — no other submission applies agentic micropayments to scientific research workflows. The ability to return early-stage hypotheses or rejected outcomes based on evidence quality (rather than always producing a final answer) shows domain awareness and responsible AI design. Overall: A top-tier submission with deep Circle + Arc integration, verified sub-cent agent-to-agent payments, 50+ testnet transactions, live hosted demo, and a genuinely novel vertical. One of the most complete and well-documented submissions in the hackathon."
Dharma Singh
Senior Development Manager
For Innovators & Creators