VelocityAI is live on Ethereum mainnet

FAQ

DEXes are a cornerstone of DeFi. DEXes solve several problems that are plaguing CEXes: custody with negligent risk management, outright fraud, and fake (wash) trading. All of these problems are not part of DEX trading — all activities are recorded in the public blockchain and thus transparent and immutable; the system is non-custodial; and DEXes are generally censorship resistant.

However, DEX trading tends to be much more expensive than trading on CEXes — both in terms of transaction costs and price impact, and therefore trading volume on CEXes dwarfs that on DEXes.

We believe that DEXes are the future of trading in crypto assets, and that DEXes could reach their potential of becoming the main trading venue for crypto assets, only when trading costs will be competitive to CEX trading costs.

We are dedicated to use our expertise in financial engineering and data science to present efficiency to DEX trading, in VelocityAI DEX and the entire DeFi ecosystem.

VelocityAI solutions address two main problems:
  • High trading costs — VelocityAI implements a novel pool architecture, which enables liquidity pools to hold, for a limited time, external tokens as reserves. By doing so, we eliminate one source of high trading costs — indirect trades. Indirect trades occur when the two traded tokens lack sufficient direct liquidity, and therefore the trade is routed via multiple liquidity pools, which results in multiple excessive costs (each pool bears its own trading fees, price impact, and gas fees).
  • Low returns to Liquidity Providers — in order to attract ample liquidity, which is required for efficient DEX trading, we strive to increase LP returns. Firstly, our novel pool architecture allows each pool on VelocityAI to serve multiple trading pairs — in principle, pool XY can serve any trade involving either asset X or asset Y, not necessarily swaps between X and Y — increasing considerably the returns to liquidity provision.

The first AI agent of VelocityAI is called Minerva — a proprietary AI-based system for optimizing protocol rewards, whose goal is to guide liquidity allocation across the ecosystem of VelocityAI pools. Minerva consists of two modules:
  • Predictive module, which takes all past trades in all pools on a given blockchain over a given period (and aggregates various legs of a given trade to estimate “desired trades”), all past liquidity provision and withdrawal events, and various other market signals — and uses AI-based models to arrive at the expected distributions of trades in all asset pairs over the next period.
  • Optimization module, which takes the predicted trade distributions as well as estimated required returns for all pairs of assets and estimated inertia in LPs’ responses to incentives as given and uses VelocityAI ’s iterative AI-based algorithm to arrive at the distribution of protocol rewards to VelocityAI pools that maximizes a linear combination of a) LPs’ expected returns net of expected impermanent loss, b) overall trading volume through VelocityAI , and c) inverse of mean trading costs on VelocityAI . This optimization relies on a proprietary heuristic method developed by VelocityAI to deal with this particular type of constrained optimization.
We are currently working on multiple additional AI agents focused on protocol-level and pool-level liquidity optimization and wallet-level trade analysis.

VelocityAI achieves the best results for asset pairs that do not have a significant liquidity pool. Most assets have a single sizable liquidity pool, typically with either ETH or a stable coin (USDC). Trading in any other pair (or: the 99% assets) is in most cases significantly cheaper on VelocityAI than on competing DEXes.
More than 20% of all trading volume on Ethereum and Polygon blockchains is composed of indirect trades involving the 99% assets.

We have done extensive simulations using real on-chain data, which provide numerical estimates of the superiority of trading on VelocityAI relative to traditional DEXes. VelocityAI allows to reduce the costs of trading by up to 50% and, at the same time, to increase returns to LPs by up to 400%.

VelocityAI has been deployed on Polygon since July 2023, and on Arbitrum since February 2024. There are expansion plans to several additional EVM-compatible chains.

There are several.
First, we use a unique, reserve-powered pool architecture to deliver superior results. Importantly, the advantages of using such architecture are orthogonal to the specific of a DEX (such as trading curve, e.g. Curve’s “stable coin math”) or particulars of liquidity provision (such as Uniswap V3’s concentrated liquidity”).
Second, our sophisticated AI based real-time optimization module (the Minerva Engine) that guides the incentivization of liquidity providers to make sure liquidity is deployed where it is needed most. We are currently developing multiple additional AI agents.
Third, and most importantly, the experience, creativity, and vision of our contributors, backers and partners, is second-to-none, as demonstrated by their past accomplishments.

$VAI is the VelocityAI token. $VAI is used for VelocityAI governance. VelocityAI tokenomics, supporting $VAI, is state-of-the-art, building on and extending current methods of ensuring token utility.
$VAI has been deployed on Ethereum and has been bridged to Polygon (where it carries the name $fxVAI) and Arbitrum ($VAI). $VAI will be bridged to every chain on which VelocityAI will be deployed in the future.

Our tokenomics is heavily influenced by Curve $veCRV tokens, used for governance. $VAI holders are incentivized to stake their tokens and receive $veVAI, used for voting on various proposals, mostly on protocol parameters and assets that can enter reserves of liquidity pools.

Unlike Curve, where you have to lock $CRV tokens to receive $veCRV, VelocityAI offers a simpler solution. By holding tokens, you can enjoy governance and cash flow benefits without the need to lock $VAI. However, you have the option to lock $VAI for additional rewards.

In addition, staking and locking $VAI increases the return to liquidity provision by boosting LP rewards with newly minted $VAI tokens. Moreover, the duration of staking (with or without locking) of $VAI tokens positively impacts the $VAI rewards to liquidity provision.

Yes, we are currently in the process of developing VelocityAI v2 and v3. The upcoming versions will incorporate the following features:
  • Multiple fee tiers; differentiated reserve allow lists;
  • Additional AMM trading functions, including concentrated liquidity;
  • Introducing multiple additional AI agents;
  • A decentralized hedging module;
  • The Initial Decentralized Offering module.

Virtus is the Roman god (or goddess) of virtue. VelocityAI is all about efficiency and fair distribution of profit in the AMM world, these are real virtues of a trading system.
Also, VelocityAI uses reserve-powered Virtual pools.