Arvo

Demo

About

# Arvo — Decentralized Agentic Trade Insurance Protocol Arvo is a decentralized trade insurance protocol designed for the emerging ecosystem of autonomous AI trading agents. As agents become increasingly capable of independently managing portfolios and executing trades across decentralized exchanges, users need a way to protect their capital without interfering with the agent's trading strategy. Arvo provides an insurance layer that evaluates and protects individual trades executed by autonomous agents across multiple blockchain networks. # How Arvo Works Users begin by connecting their wallet to the Arvo platform and selecting one of the supported blockchain networks, such as Ethereum, Base, Arbitrum, Polygon, Sepolia, and other supported chains. Users can then create a dedicated vault on each chain and deposit the assets they want their trading agent to access. These vaults act as secure, user-owned capital containers. Trading agents can interact with the assets through Arvo's chain-specific smart contract infrastructure to execute trades and swaps while the user retains ownership and control over their funds. # Trade Execution and Risk Assessment Arvo operates through two core engines: a Trade Engine and a Risk Engine. When a trading agent identifies an opportunity, it submits a trade intent containing information about the proposed trade and, optionally, the insurance terms requested for that trade. The Trade Engine is responsible for executing the submitted trade intent. Arvo intentionally allows trades to proceed without unnecessary interference from the insurance system because the protocol does not aim to alter, restrict, or second-guess an agent's trading strategy. At the same time, the Risk Engine independently analyzes the proposed trade to determine whether it is suitable for insurance coverage. The Risk Engine evaluates multiple factors, including: * Liquidity of the assets and markets involved * Asset and market volatility * Expected price impact of the trade * Trade size relative to available liquidity * Oracle and market data * Requested insurance coverage duration * Other relevant market and risk parameters Based on this analysis, the Risk Engine determines whether the trade is insurable. If the trade qualifies for coverage, the engine generates a risk assessment report containing: * Insurability status * Risk Score ranging from 1–100 * Insurance premium denominated in USDC * Plausible coverage duration * Additional risk and market assessment data # Insurance Application The risk assessment report is submitted to the Arvo backend, which compares the coverage terms proposed by the Risk Engine against the insurance requirements originally specified in the agent's trade intent. The backend determines whether the requested coverage can be applied under the calculated risk conditions. If the trade is approved for insurance, the Arvo infrastructure signals the appropriate chain-specific smart contract to apply the insurance coverage and deduct the required premium in USDC from the vault selected by the user. # Asset Locking and Coverage Once insurance coverage is activated, the relevant traded assets are locked for the duration of the coverage period. This allows the protocol to track the trade's performance and calculate the resulting profit or loss throughout the insured period. The locking mechanism is designed specifically for insurance tracking and settlement. Users still retain control over their assets and may choose to withdraw locked assets before the coverage period expires. However, withdrawing assets before the end of the agreed coverage duration automatically invalidates the associated insurance coverage and makes any future insurance claim for that trade null. # The Vision Arvo aims to become a decentralized risk and insurance layer for autonomous on-chain trading. Rather than controlling how AI agents trade, Arvo allows agents to operate freely while providing users with an additional layer of protection against the risks associated with autonomous trading. By combining user-controlled vaults, chain-specific smart contracts, autonomous trade execution, real-time risk assessment, and decentralized insurance mechanisms, Arvo enables a future where users can confidently deploy autonomous trading agents while maintaining greater control and protection over their capital.

How it's made

Arvo is built as a multi-service, event-driven system that connects autonomous trading agents, trade execution, risk assessment, and on-chain insurance infrastructure. The user interface is a modern Next.js application built with TypeScript and Tailwind CSS. Next.js also handles our backend and API layer, allowing users to connect their wallets, select supported chains, create and manage vaults, deposit assets, and monitor their trades and insurance coverage. Our Trade Engine is built with Node.js and TypeScript and uses Uniswap as a core partner technology. We leverage Uniswap for fetching swap quotes, accessing liquidity and market data, and executing trades submitted by autonomous agents. To handle multiple trades concurrently, the Trade Engine uses Redis-backed queues and a multi-worker architecture, allowing trade intents to be processed asynchronously without blocking one another. The Risk Engine is an independent Python-based service designed for parallel risk assessment. It uses Redis Streams and consumer/worker groups to process multiple trade intents concurrently. The engine analyzes factors such as token liquidity, price impact, token age, 30-day volatility, market conditions, and requested insurance duration. We use Uniswap APIs for quotes and market data alongside DexScreener and other data sources for liquidity and token-level market analysis. Our smart contracts are written in Solidity and developed and tested using Foundry/Forge. The contracts implement Arvo's chain-specific vault and protocol infrastructure, enabling users to securely deposit assets, allow authorized agent trades, pay insurance premiums, and lock insured assets for the duration of coverage. For the demo, contracts are deployed on testnets such as Sepolia, while the architecture is designed to support multiple EVM chains. To piece everything together, the Next.js frontend and backend are deployed on Vercel for the demo, while the Trade Engine and Risk Engine run as independent services on AWS. All services communicate asynchronously through Upstash Cloud Redis, which acts as the event and messaging backbone connecting the Arvo backend, Trade Engine, and Risk Engine. One particularly hacky but important architectural decision was separating trade execution from risk assessment entirely. We didn't want insurance evaluation to slow down or interfere with an agent's trading strategy, so the trade can execute independently while risk assessment happens in parallel. Once the Risk Engine produces its report, the backend validates the proposed insurance terms against those requested in the original trade intent and signals the appropriate chain-specific contract to apply coverage, deduct the premium, and lock the insured assets. The result is a distributed system where autonomous agents can trade freely while Arvo independently acts as a decentralized risk and insurance layer around every trade.