Crypto AI Agents Are the Digital Minions of the Blockchain

UPDATED Jan 3rd, 2025

Wouldn’t it be nice to have little yellow minions running around doing your bidding? While we may be many years away from physical minions, the digital ones have arrived: crypto AI agents! And we expect them to take over lots of tasks already in 2025.  

AI agents are, put simply, bots trained on large language models that can handle complex tasks, either alone or in harmony. Luckily, they’re a lot smarter than the minions we’ve grown to love from the film Despicable Me. They won’t accidentally blow up your house or drive your car into a hot lava pit.

While an AI agent can’t do your laundry, they can handle large-scale project management, community organization, market analysis, and more. When combined with the blockchain, these fancy bots become powerful tools for scaling businesses and personal crypto goals. You can learn more about the combined use cases of AI in blockchain from our deep-dive report.

Since our initial publishing on 11 July, they’ve come a long way. It may seem like AI is everywhere thanks to the AI hype that started earlier this year. Below you’ll find our original article plus the latest hot takes on AI’s advancement in Web3. First, let’s review exactly what AI agents are.

What is an AI agent?

You probably already know what an AI agent is, but here are the main characteristics.

AI agents are trained on large language models and mature over time using the data they receive from various inputs, including human input, microphones, cameras, and radar. Luckily, there are plenty ready for use out of the box. These no-code solutions drive up access to both AI automation and the power of the blockchain, so you don’t have to learn how to build an AI agent yourself. Hoorah!

So, while you may be wondering ‘what is an AI agent?’, you’ve probably used one already. The robot vacuum your cat rides around the house utilizes an AI agent to develop a spatial map of the rooms. Your smart thermostat uses a rigid framework to determine the appropriate reaction to if/then statements to keep you cool in summer. Autonomous vehicles use multi-layer AI agents to sort through mass inputs of data to make sure it doesn’t run a red light (or over a dog). And yes, ChatGPT is a very popular example of an AI agent.

What’s more, AI agents are reaching ubiquity in crypto. Most projects are exploring some kind of AI integration, though you the user may not realize it. AI agent integration is looking especially hot in DeFi. You’ll get an update on the top blockchain AI projects in a bit. Taking a step back, there remain some issues with AI that should be kept in mind.

Limitations of the AI agent economy

There are a few key issues that currently exist in the AI agent economy that you should be aware of (beyond your mom not trusting them). 

1. I’m not bad; I’m just drawn that way.
   Data privacy concerns persist when it comes to AI. In June 2024, Meta rolled out a new line in their terms and conditions that states their AI will be trained on all Facebook data (unless opted out). This caused quite the stir for its users, and influencers across the web were quick to urge followers to opt out. So much so, Meta halted the plan in the EU. In August 2024, the EU AI Act began its rollout, working to ensure greater privacy and transparency in AI.
2. I don’t think we’re in Kansas anymore…
   Or anywhere else in reality. It’s true that ChatGPT sometimes makes up its own data. Often enough, the stories it tells develop from confirmation bias fed by its user (the human). This can make it hard to trust AI to execute well on mission-critical projects. While it is believed that hallucinations are likely to continue, engineers have been working to provide better data and stricter parameters to AI to reduce the risk.
3. When someone asks you if you’re a god, you say YES!
   AI isn’t all-powerful, and it still can be hacked. It might want you to think it’s a god, but it’s not. It’s just super-fast computing that can access vast amounts of data. Like all computers, AI computing can be hacked. This remains a critical issue in December of 2024, with 32% of AI projects experiencing accidental exposure to vulnerabilities, 50% of which were significant.

But never fear — the blockchain is here! And with it, crypto AI agents.

What happens when we combine AI agents and Blockchain? Magic… 

Arthur C. Clarke said, “Any sufficiently advanced technology is indistinguishable from magic.” That’s how we feel about crypto AI agents. When combined, AI agents and blockchain technology solve the above problems and provide new scalability to blockchain networks. Let’s unpack the beauty of crypto AI agents:

• The blockchain is inherently open-source, meaning the data is free and available for AI agents to train on. That data is also reviewable, ensuring mistakes are found and corrected. Projects like SubQuery are providing AI models with such open source data. We like that!
• A single crypto AI agent might incorrectly decide that market movement is terrible and try to sell all your tokens. A swarm of crypto AI agents will check each other — if one goes rogue, the others will verify against the Rogue One (Star Wars reference intended) and eliminate the issue. New projects like IQ are making it even easier for AI agent clusters to be deployed. This is why we are fans of decentralized AI for our projects. 
• The blockchain is famously difficult to hack. Each chain is verified by multiple nodes. Just as a cluster of crypto AI agents verifies against each other to ensure correct action, any node on the chain that gets hacked will be noticed by the others because it no longer matches the data held by all. When crypto AI is active on the blockchain, their security increases enormously. This is happy news for anyone investing in web3 ai tokens.

But wait, there’s more! AI in crypto has additional benefits that improve blockchain utility and accessibility. Let’s check those out.

• Trading becomes simple — someone who doesn’t have time or know-how can use a crypto AI agent to watch the markets and make decisions for them. Add to this the speed AI agents bring to crypto, and this becomes one hot option. We think AI trading bots are going to be hot in 2025. The emergence of AI crypto trading bots with their own personalities has been an additionally fascinating trend.
• Blockchain transactions can get pricey (ETH gas, anyone?). A cluster of crypto AI agents can watch gas fees and make decisions according to set thresholds. The Artela Network has been working on this since August.
• Naturally, AI in web3 applies to smart contracts. Smart contracts are predetermined triggers on the blockchain. The creation of the smart contract enabled NFTs (non-fungible or individually unique tokens) to work on the blockchain. A smart contract boils down to a list of rules. When applied to a crypto AI agent group, it delivers a fabulous level of automation. This has awesome implications for DAO (decentralized autonomous organization) projects, as it can automate the management of a large group of members and their goals. Phala Network has been leading the way on this front in the latter half of 2024.

The AI Agent Crypto Projects we’ve fallen in love with

If you’re anything like us, you need to see examples of Web3 AI projects being used in the wild. Here are the AI agent crypto projects we can’t stop talking about because they leverage both large and small scale applications.

• Are you a business owner who wants to leverage the tech but doesn’t want to learn how to build an AI agent? Fetch.ai provides a plug-and-play solution to help businesses integrate them without an overhaul to code. They even tout an upcoming AI Agent economy, a mesh of AI Agents that work together to accelerate business activities across Web3. PS — Fetch.ai is a member of the Artificial Superintelligence Alliance which just rolled out its end of year update.
• What if you could have Warren Buffet do all your crypto trading? He’s probably unavailable, but a crypto AI Agent can absolutely be trained on his trading history and apply it to current crypto markets. In fact, Autonolas just deployed a community-owned AI Agent doing just that!
• Or perhaps you want to let crypto AI agents do all the work for you. Harambe, an AI-powered token that launched mid-2024, is designed to trade itself to grow your portfolio. And yes, the logo is a gorilla.
• You may have already played with AI generated art. But have you considered what digital art might look like if everyone in a DAO had input in the creation of a single piece? That’s what Botto is doing with their crypto AI agents, and the results are trippy.
• Or maybe you simply want to make your NFT talk, then you’ll opt for talkens.ai.

Most of this article was written already in July 2024. The trend has only intensified and there’s no sign of it cooling down. On the contrary, we expect crypto AI agents to have a realistic chance of becoming as popular as the minions – if not more. Onchain’s Head of Research, Leon Wadimann will share his predictions and the reasons behind them in the upcoming report Web3 Predictions 2025.