15 Crypto AI Projects You Should Know About
Table of Contents
AI tokens like FET, AGIX, and ALII grew 400% and beyond in early 2023, while Worldcoin took center stage in Q3. It’s hard to keep up with every new project, so here’s our detailed guide to AI in blockchain - especially timely since Pontem is working on an AI product of its own!
Is Blockchain a Good Match for AI?
The beginning of 2023 was marked by a huge rally in AI tokens, with assets like AGIX and ALI growing 1000%. The rally had a clear trigger: the success of ChatGPT, a natural language processing model developed by OpenAI. There was another short-lived rally in July, after Elon Musk announced the creation of xAI, a potential competitor to ChatGPT.
There is little doubt that AI will revolutionize many industries. Fortune Business Insights forecasts that the AI market will grow by 20.1% annually to reach $1.39 trillion by 2029. However, the question remains if AI needs blockchain, and vice versa.
View no.1: AI & blockchain is just a narrative
Andre Cronje, the creator of Yearn Finance, wrote that “AI and blockchains do not mix.” He even went so far as to say that crypto projects that “jump on the AI bandwagon” do so only to pump their tokens.
If Cronje is right, then the AI narrative in crypto will fade away like many narratives before it. For example, during the first ICO boom in 2017-2018, many claimed that blockchain would “disrupt” healthcare and supply chains but that hasn’t happened yet.
Pratik Thakar, head of creative strategy at Coca-Cola, recently advised marketing executives to invest in AI rather than blockchain, as the former will be “more fundamental”.
To be fair, many of the AI blockchain projects that we examine in this article were founded in 2017 and 2018, long before ChatGPT. It would be unfair to say that they jumped on the bandwagon as AI became a household name.
View no.2: Blockchain is needed to verify data
Contrary to Andre Cronje, Aptive Resources Program Director Ralph Wallace believes that blockchain can solve the issue of data integrity and trust in AI, by verifying data sources. According to him, if the data is low-quality, you’ll get “garbage in, garbage out” as a result.
View no.3: Blockchain as a way to detect AI-generated content
Kamlesh Nagware, co-chair at Hyperledger India writes that blockchain can become “the trust layer” for the AI industry, verifying the authorship and origin of a piece of content, rather than the provenance of training data. We’ve actually discussed this topic during the recent Move Tuesday.
This use case could become especially important as more future AI models on content generated by other AI algorithms. Research shows that this feedback loop leads to a decline in quality and even “model collapse”.
View no.4: Web3 products that incorporate generative AI are the way forward
The easiest way to incorporate AI into Web3 here and now is to integrate generative AI, especially large language models (like the ChatGPT API) into blockchain products like wallets and exchanges For example, imagine a crypto wallet that can answer questions about the tokens it supports, or a DEX that can provide basic technical analysis insights.
Pontem Network is actually working on something in this vein. We won’t show our hand just yet, but keep an eye out for our beta launch soon!.
The use of generative AI could even go deeper than dApps. IntoTheBlock CEO Jesus Rodriguez writes that AI-generated natural language could be used in blockchain transactions and smart contracts.
It’s impossible to predict which of these views will turn out to be correct In any case, it’s important to get familiar with this section of the space and its core projects.
AI Blockchain Projects At A Glance
AI-related blockchain projects can be categorized by their core features::
- Services provided by AI ( processed data, market predictions, etc.);
- AI algorithms (image generation, language processing, virtual assistants, etc.);
- Data needed for training AI models (though not an AI product proper, it’s a crucial resource for developing ML models. In January and February 2023, Big Data tokens pumped together with AI tokens);
- Computing resources for training AI models;
- Secondary products,, such as generative AI characters or DeFi..
(Note that we’ve included only those projects that directly address the AI industry and have a market cap above $2 million. That’s why Big Data and decentralized computing protocols like The Graph and Render Network are omitted.)
The Top 16 Crypto AI Projects You Should Know About
Pontem Network’s mystery AI blockchain product - coming very soon!
For the past couple of months, we’ve been working on a very useful new feature - and yes, it’s powered by a large language model. We won’t reveal the details yet, but you’ll be able to join beta testing very soon!
Worldcoin is a controversial project headed by Sam Altman, CEO of OpenAI, the company behind ChatGPT. The idea is that soon the internet will be so full of AI bots that we’ll need a way to prove that we are human. For this purpose, Worldcoin proposes “World ID” - a digital passport based on your iris.
To get verified and create a World ID, a user needs to have their eyeball scanned by a silver ball called an “orb”. Only verified users will be able to use WLD tokens and the corresponding payment app. World IDs will be stored on the blockchain, and the creators insist that users’ privacy will be protected.
The 18-month beta testing period attracted over 2 million users. On July 24, 2023, Worldcoin finally launched: orb verification should soon expand to 20 countries, bringing the number of potential registrations to 200,000 a week. On the same day, WLD token started trading on Binance.
As the CEO of one of the most famous AI startups in the world, Sam Altman understands the risks of AI and insists we will need a new way to distinguish humans from bots. However, Worldcoin’s practices have raised concerns: Vitalik Buterin himself writes that privacy risks are too great with biometric proof of identity (as an eyeball scan can reveal a lot about a person). Plus, he notes, it will be difficult to get orbs to everyone who wants a World ID, and some governments might even decide to ban the technology altogether.
SingularityNET envisions a future where decentralized AI performs many of our daily tasks. The project uses the term “Artificial General Intelligence” (AGI) to show that neural networks will be versatile, rather than specific to just one job. The token symbol AGIX, comes from AGI.
Their key difference from Fetch.ai (which we’ll look at next) is that instead of monetizing the services performed by AI agents, users can monetize AI models themselves. A developer can train a new machine learning algorithm, upload it to SingularityNET using a special Publisher tool, then sell it on the marketplace. Customers pay with AGIX or fiat.
The marketplace already features a few free AI demos. They include an image generator and an algorithm that generates music lyrics based on a seed you type into it.
Fetch.ai is a decentralized AI blockchain platform centered on the concept of Autonomous Economic Agents (AEAs): AI algorithms that perform tasks on their own, without human inputs.AEAs can make decisions, communicate with each other, and improve at their job over time.
Fetch.ai envisions a future Internet of Things (IoT) as a multi-agent system of AI workers who do their job as instructed by their “masters”. Users can then monetize the results of their AEAs’ work.
Autonomous agents might gather and process data generated by various sensors and thus power smart energy grids and whole smart cities. FET tokens would be used to pay for their services. Other use cases include crypto market analytics and social networks.
The FET token “lives” both on Ethereum as an ERC-20 and on the Fetch.ai mainnet. You’ll need Fetch Wallet to use the native token, as this blockchain isn’t supported by MetaMask.
Ocean isn’t an AI protocol in itself, but it deals with the thing neural networks need most: data. Ocean is building a decentralized marketplace for data where providers and customers are matched without intermediaries.
Ocean uses special ERC-20 datatokens and ERC-721 NFTs as access keys for datasets. Customers need to pay with datatokens when buying datasets. Data itself can be stored anywhere, including centralized servers, but access to it is decentralized.
Users can lock OCEAN to earn around 6.5% APY and receive special veOCEAN tokens in return. If you also allocate your veOCEAN to participate in the curation of certain data, you’ll get additional rewards.
Currently, the marketplace mostly features crypto trading data, such as ETH/USDT order books and CEX aggregators.
Alethea uses generative AI to create characters based on a text description. You can describe the desired appearance, personality, profession, skills, and so forth. The technology is called CharacterGPT.
Every virtual character is minted as an iNFT, which consists of three parts: soul, mind, and body. iNFTs are dynamic;users can add new data to them to change the way the virtual character behaves or looks. The characters can even blink, nod, and do other movements. They also learn, like any neural network. Very interestingly, you can turn a regular NFT (like a Bored Ape) into an iNFT.
You can try CharacterGPT in action with the MyCharacter.ai dApp, which allows you to create 5 characters for free. We entered “A Space Pirate who loves crypto and swashbuckling”, but for some reason the result was a friendly young baker – a far cry from the real Pontem Space Pirate NFTs!
Another innovation of Alethea is the AI Protocol. It allows developers to integrate CharacterGPT and iNFTs into their own projects. Users need ALI tokens to create and upgrade iNFTs, or to interact with the AI Protocol. The token also powers the DAO, whose members elect the Content Moderation Council.
Numerai is a combination of a hedge fund and a community of data scientists. It trades using crowd-sourced AI algorithms to predict crypto prices. Outside investors can put money in the hedge fund to hopefully enjoy the returns.
In economics, a numeraire is a unit used to compare the prices of different assets. USD is the traditional numeraire for stocks. Combine “numeraire” and “AI”, and you get Numerai.
At the heart of Numerai is a system of tournaments in which community members compete to create the best machine learning algorithms. They stake NMR, and if one’s AI model does a good job predicting prices, the author is rewarded with more NMR. But if the algorithm performs poorly, the user’s stake is slashed.
This motivates data scientists to submit better models, which Numerai combines into a weighted meta model. Considering that some users earn over 300% APY in NMR thanks to their algorithms (crowdsourcing the leaderboard), crowdsourcing AI is working so far .
Cortex is positioned as a decentralized computer for AI dApps. The challenge with running AI models on a blockchain is ito achieve deterministic inference an algorithm which produces the same result based on the same input regardless of the environment. Cortex purports to be the first L1 blockchain to have the necessary characteristics for decentralized AI.
The project features an EVM-compatible virtual machine (CVM) and supports Solidity. Cortex also has its own ZK (zero-knowledge proof) rollup called zkMatrix. It should help scale the network and reduce fees by processing many transactions on a sidechain.
Cortex introduced a system of rewards for independent AI developers who add their algorithms to the ecosystem. dApp developers will be able to select those that better match their use case. There is already an AI dApp for you to try: DigitalClash, a game where users compete to draw pixels on a board.
Phoenix is a new blockchain optimized for Web3 AI dApps and IoT. It includes an EVM-compatible Layer 1 (mainnet is live with around 1,000 transactions per day) and a Layer 2 (computation layer) that will host applications. You can stake PHB to earn around 17% APY.
The project proposes to connect on-chain applications with off-chain datasets, models, and machine learning frameworks. For that, Phoenix has a built-in oracle, though it currently only supplies simple data like crypto and stock prices, Covid-19 cases, and the temperature in Hong Kong.
SingularityDAO is an AI-powered DeFi platform built by a team of machine learning scientists, data engineers, and quant trading experts. It is built and incubated by the SingularityNet Foundation.
SingularityDAO’s key innovation are DynaSets: portfolios of tokens created and automatically rebalanced using AI. In other words, they are trading strategies controlled by an AI-powered Dynamic Asset Manager.
The two DynaSets currently on offer are dynBTC (WBTC & USDC) and dynETH (WETH & USDC). Users can also test several new beta strategies in the DynaLab section.
DeFi functionality also includes a swapping DEXand vaults and farms where you can earn over 10% APY by staking tokens.
Vector Space AI is a subsidiary of Vector Space Biosciences, which hopes to use machine learning to send humans into space. In order to reach Mars, astronauts will need special systems to protect them from radiation and the effects of very low gravity in the spaceship.
Vector Space’s idea is to use AI to construct datasets and reveal hidden relationships between genes, molecules, etc. This should speed up breakthroughs in space bioscience. Interestingly, Vectorspace applies the same AI technology to uncover connections between the prices of cryptocurrencies, global events, and other factors to build whole portfolio strategies.
The project uses blockchain to verify the provenance of data in each dataset. Institutional customers require VXV tokens to pay for access to datasets.
VAIOT creates AI-powered mobile virtual assistants. They can be used by businesses to sell products like insurance to customers, prepare personalized offers, and even draw up legal contracts.
The VAIOT assistant uses a natural language model, so communicating with it is more pleasant than with a standard support bot. Unlike bots, AI assistants can also finalize offers, concluding sales on their own. In the process, they gather valuable information that can help improve business processes.
The first real-world product - a solution for social media campaigns with rewards - should be released soon. Transactions are recorded on VAIOT’s own blockchain, and VAI tokens are needed to pay for services and receive payback (cashback). You can also farm VAI on Uniswap.
GNY is building an AI prediction platform for crypto trading. The first product is the BTC Range Report - an AI-generated prediction for the Bitcoin price range for the next 7 days, updated daily.
A machine learning algorithm analyzes millions of trades to find patterns, though the range is admittedly wide: between $21,500 and $28,000, for example.
Access to the Range Report costs $20 a month, and Ethereum has just been added. In the next few months, GNY plans to expand to 10 cryptocurrencies.
GNY has its own blockchain, where the GNY token is used to pay for AI products, launching new assets, voting, etc. There’s also a mainnet wallet. GNY is available on Ethereum and BNB Chain, as well.
DeepBrain Chain is an AI computing network: nodes across the globe provide their computer resources to AI developers. Training a neural network takes a lot of computing power and costs AI projects a lot of money. But with DeepBrain Chain, they can outsource this resource-intensive task to others in a decentralized way and save up to 30% (according to the website).
The platform serves as a matchmaker between AI developers and nodes running GPUs (graphic cards). These range from big GPU rigs to regular users with a single Nvidia card. GPU owners can thus consider DeepBrain Chain as an alternative to crypto mining.
DeepBrain Chain supports major machine learning software frameworks such as TensorFlow,
DeepBrain Chain is already powering AI applications in cancer detection, driverless cars, and speech and image recognition. Several cloud computing providers based in South Korea and Singapore provide GPU services to users, with over 2,400 GPUs in the network.
Oraichain is yet another Layer 1 blockchain built for decentralized AI, as well as Web3 in general. It uses Cosmos SDK and CosmWasm, but it also has a subnetwork for AI dApps that is EVM-compatible. There is also a special trustless oracle for AI, as well as a ZK rollup for scaling.
The native OWallet combines the features of the Kepler wallet for Cosmos and MetaMask. It supports Oraichain, Cosmos, Osmosis, and Juno, together with Ethereum, BNB Chain, and other networks. You can also add the Oraichain EVM subnetwork to MetaMask.
There are 10 dApps in Oraichain’s ecosystem, including DINO Hub, a marketplace for AI APIs and data. It supports royalties, so AI developers can earn regular, passive income. For example, you’ll find AI services for weather forecasting, detecting heart diseases, and even de-aging people in photos.
Matrix is a blockchain and an AI cloud computing platform. It uses a combined Proof of Work + Decentralized Proof of Stake consensus mechanism where nodes that aren’t directly involved in mining or validation can lease their computing power to others. This makes mining environmentally friendlier and allows the network to process up to 14,000 TPS.
Other features include:
- MANAS, a marketplace for AI services like logo recognition and plant identification;
- MANTA, an automated machine learning platform;
- MANIA, a tool for turning AI models into tradable NFTs;
- Airtist, an AI protocol that generates NFT artwork.
The name “Matrix AI” points to the team’s intention to bring the vision of the famous Matrix films to life, including intelligent avatars. Starting in 2023, the project will begin upgrading to Matrix 3.0, which would integrate neural data with AI and blockchain for the first time. Users will even be able to upload their brain wave patterns to the blockchain to power their personal avatars in the metaverse.
Raven is a decentralized computing protocol for AI, metaverse, gaming, and Web3. It aims to connect millions of devices that have spare computing power with those who need them (requesters), such as machine learning engineers and game developers.
Developers can use the Ravenverse app with its various libraries to build and train AI models and even create games. Various parts of the Ravenverse allow users to assign computation tasks to free nodes as a requester, participate in current computations as a provider, etc. All this requires RAVEN tokens.
Crypto AI projects on Aptos
So far, the majority of AI projects have either built their own blockchains or issued ERC-20 tokens. However, as Aptos has the superior speed and scalability needed for machine learning platforms, we will likely see more AI projects launch in the Aptos ecosystem soon.
There are a couple more projects we can add to the list:
- AptosLauncher: a launchpad on Aptos that plans to use machine learning technology created by OpenAI.
- Aptos Robi: an NFT project run by artificial intelligence. The AI created designs of 444 robots that can develop and interact, and most of the texts for the project are also written by an algorithm.
We’ve already mentioned that Pontem Network is working on an AI product as well, so stay tuned!.
Decentralized AI is a very interesting trend that can turn out to be more than just a short-lived narrative. We will keep reporting on this space, so don’t forget to follow Pontem on Telegram, Twitter, and Discord so that you don’t miss the next article!
Pontem Network is a product studio building the first-ever suite of foundational dApps for Aptos. Pontem Wallet, the first wallet for Aptos, is available for Chrome, Mozilla Firefox, Android, and iOS.
Use Pontem Wallet to store and send any tokens on Aptos. The wallet is integrated with our Liquidswap DEX, the first DEX and AMM for Aptos, Topaz and Souffl3 NFT marketplaces, Ditto and Tortuga liquid staking platforms, Argo and Aries lending protocols, and all other major Aptos dApps.
Our other products include the browser code editor Move Playground, the Move IntelliJ IDE plugin for developers, and the Solidity-to-Move code translator ByteBabel – the first ever implementation of the Ethereum Virtual Machine for Aptos.