AI is what takes care of mundane tasks in crypto. There’s actually plenty of ways how AI is useful in crypto, most of which aren’t as obvious as might seem. For starters, AI agents can launch their own crypto and what’s even more interesting — CEX’es use AI each day to help you trade better. Let’s take a deeper dive into what use AI has in crypto.
Since the launch of OpenAI’s ChatGPT in 2022 generative AI’s and LLM’s have become the new «Next Big Thing»: healthcare, insurance, big-data sciences and supply-chains suddenly got a power up. But is there any use for AI in Crypto?
AI is the first «Face Control» you’ll encounter in crypto. Each time you open up a Crypto Trading Account every KYC you pass is AI-assisted face recognition. Major CEX’es, like OKX, Binance, Kraken use AI’s to prevent drainage of a trading account or breaches based on what that account does. When you send or receive crypto, every AML check powered by Chaynalysis Reactor — a mainstream software for compliance and cybercrime investigation — is reinforced by AI to filter out dirty crypto and risky connections. While it’s tedious to complete multiple KYC’s, with NonBank it’s a one-time deal — set-up NonID and you’re in.
AI is essential to keep crypto running smoothly. For instance, you have AMM’s — or automated market-making protocols, which are run not by people, but by a sort of AI. AMM’s essential to crypto because they provide funds for all the trading that goes on in the market. Think of AMM like pockets with money everybody leans into when they place orders to swap tokens via Uniswap or any other DEX platform. Without AMM’s you would have to wait for someone to login and sell you crypto manually.
Here's the top 5 coins, run by AMM: UNI, JUP, AERO, RAY and CAKE — every one of them successfully surpassed the $1 price threshold and made a name for themselves. For instance, CAKE represents an ecosystem of PancakeSwap apps, where people can launch, exchange and convert cryptos. Uniswap, or UNI, is the workhorse of the AMM market, boasting ~$4,7M Market Cap alone, known for its low fees and huge retrodrops. Raydium is the breeding ground of PUMP.FUN memes — an automated memecoin kickstarter on Solana.
The market for AMM’s is big, with a total cap of ~$35B and daily volume of ~$6M — that’s a big number, much like total NFL bets in 2024 season or exactly how much Bill Hwang burned to the ground trying to «corner the market» while being corned with fraud accusations.
AI is used by MEV Bots, who gather Maximum Extractable Value — revenue from shuffling transaction order inside the blocks the very instant they are about to get recorded on-chain. Thing is that after PoS becoming widely adopted, blocks are produced by multiple nodes, which work to sustain on-chain activity. Each block produced in the PoS chain contains a number of transactions within — even yours — that has to be recorded. Some transactions are cheaper to embed, some are not, so MEV bots utilise AI to find that profitable notch, and essentially front-run a transaction execution.
Here’s how MEV works:
Here’s how big a MEV Market;
Here some places to research MEV Data:
AI is what powers Data Oracles. So, you’ve got a huge set of smart-contracts and they need external data to operate — prices, volume, orderbook data and such. Data Oracle is a smart-contract entity that reads data off websites, APIs, RSS Feeds, LP Pools, Blockchain Explorers, sniffs out On-chain Data and sends it back to your huge set of smart-contracts. Data has to be relevant, accurate and helpful; otherwise it won’t cut it, so, you want someone to double-check the big data — like an AI Agent, who can see through the layered nature of the market with ease. The AI would scan each layer of data independently, looking for useful notches. This allows DEX platforms to provide the most accurate prices, helps LP Pool owners to mitigate arbitrage leaks when they are pestered by MEV bots and to account for impermanent loss when calculating APY — or profit returns — of the pool.
Most of the automated trading is done by AI’s. There’s algo-trading strategies, which use Machine Learning and Reinforced Machine Learning with a worldwide market size of $17B, including stocks and Trad-Fi stuff like Government Obligations. Those algo-models are easily exposed to the Flash Crashes — a black swan events that wipe out big parts of the market in seconds, so, naturally, they’d want to use AI to determine the risk of such an event on the fly. Crypto is the same part of the ML Trading market, but its portion is much smaller compared to stocks, yet. If you want to know just how much — there’s a report by IMARC Group on algo-trading, but it’s behind a paywall.
AI can do smart-contract audits for you, like with etherscan gpt4o integration, and help to avoid hidden honeypots. Certik’s Skynet can help you evaluate risks and check just how many audits by third parties the project completed during its lifetime.
AI can run meme accounts on X and help to promote its own coin. There’s a story from 2023 about a Goatseus Maximus coin, the successor to the Goatse meme, which grew 1200% overnight seemingly out of nowhere. The source of the signal was from X.com (let’s face it — we all call it Twitter) called @truth_terminal, which got a 50k grant from a16z co-funded — Marc Anderssen. Shortly after receiving the grant, @truth_terminal announced a GOAT token launch. Word «Goatseus» comes as successor in reference to Greek Heroes — Perseus, Theseus — and «Maximus» being «greatest of» postfix. Therefore, the name can be translated to an «The All Goat» or «He, the Goatest».The GOAT price started at $0.00005, and now it’s trading around $0.5 — an x10000 increase, sounds good, right? Trick is that only early birds, who supported the coin launch on pump.fun, got the GOAT token before it was deployed to Raydium — aka, hit the main markets.
Is there a blockchain you can use to run AI? A small heap of: Internet Computer, Render Facilities (RNDR), BitTensor and The Graph (GRT). Each provides a decentralised virtual environment of sorts that suits AI training & development needs. For instance, «Internet Computer» offers NVME storage and high-speed bandwidth. The decentralised computing market is as old as the blockchains themselves, since every miner or node is a part of a global decentralised computing network, like Ethereum. Total market size of the Decentralised Computing Market would be a sum of all the profit and transaction volume generated by PoW miners and PoS nodes from 2009 to 2024, paid out in BTC, ETH and all the PoW and PoS Cryptos out there. That’s around 4,927,500 BTC and 2,300,000 ETH only for two major chains alone; there's also Tron growing by 30% quarter-over-quarter in 2024, which is entirely based on PoS payouts.
3 things: black box, spontaneous synchronisation and single-minded crowd issues. When so many AI Agents are exposed to the market at same time, they might learn from one another, adapt to common strategies used by their counterparts. In a sense, you can look at it, as a giant terrarium, filled with different reptiles. At some point, they’ll form an ecosystem of relationships, and every iguana, snake and lizard will find their role within the scheme. If loop from outside, this closed ecosystem would be highly vulnerable for things that are not encountered inside the terrarium. This may set up a chain for a sequence of events that would trigger every AI Agent in a way that may be exploited by third parties.
Good news — manipulating the crowd of AI Agents would be insanely hard, even harder than Doom Ethernal on Ultra-Nightmare for a person who has never played FPS games before. However, the law of spontaneous synchronisation in the physics of pedestrian bridges is a good example of what can happen when the market of more than 51% AI Agents is triggered in a wrong way. In short — when pedestrians are walking past, they make it wobble, so you need to account for that, because it can break the bridge.
Thanks for reading, don’t forget to try the NonBank App and try out NonID Web3 Auth tool.
