Experts: AI Agents Will ‘Supercharge’ Crypto in 2025
- Experts say that AI will interact more with crypto and enhance the functionality of blockchain tech in 2025.
- They believe AI-powered solutions at the application level, rather than AI tokens, will be the real game-changer in this space.
- In DeFi, AI will act as “both a passive help and an active participant.”
By design, the blockchain would free machines from humans’ central authority. Now, we are witnessing the birth of AI systems with unprecedented decision-making capabilities, taking on roles traditionally managed by people.
The main actors of the coming supercycle are known as AI agents, autonomous programs designed to perform specific tasks. Thanks to blockchain, these agents can operate within permissionless networks, enabling them to autonomously make decisions, enter deals, tailor interactions, and represent companies independently.
As AI continues to develop at breakneck speed, it might interact more with crypto and enhance the functionality of blockchain technologies in 2025. Pundits believe that AI-powered solutions at the application level, rather than AI tokens, will be the real game-changer in this space.
This integration could also reshape the nature of AI itself. Some experts argue that these agents are beginning to exhibit “agency,” a characteristic once thought to be uniquely human. This fusion of AI and blockchain could mark a pivotal moment in how technology operates and interacts with the world.
Speaking to Cryptonews, Luis Bezzenberger, product development lead at blockchain infrastructure company Brainbot GmbH, said:
“The blockchain has the potential to ‘set free’ AI by giving full autonomy and agency to large-language model (LLM)-powered agents through blockchain wallets, enabling these agents to transact like humans. Ultimately, this could lead to what might one day be considered ‘synthetic life,’ where AI operates autonomously within financial and other ecosystems.”
A More Life-Like Nuance
AI is often criticized for reducing complex human interactions to algorithmic processes. For example, credit companies use algorithms to evaluate a client’s creditworthiness by analyzing data from multiple sources. Search engines and social media companies also use algorithms to tailor content for particular audiences.
The idea behind autonomous AI agents is that to convincely act on our behalf, machines don’t need to imitate humans—they just need to be better at being machines.
Utility-driven AI agents include Aixbt by Virtuals, which provides sophisticated investment research, and Zerebro, which produces unique digital art. These agents have access to far more data than traditional chatbots to analyze for both business and cultural advantage.
Designed to evolve continuously, these agents not only improve at their current tasks but also gain the ability to handle new ones and adapt to greater levels of complexity over time. This could make them increasingly reliable for future assignments.
“AI-driven systems could enable highly specific and custom-hedging strategies, such as hedging against the default of a specific company where no credit default swaps or insurance options exist,” Bezzenberger said. He added:
“With just a prompt, an AI agent could structure a bespoke options or derivatives strategy using liquid derivatives as proxies. Intent-driven transaction infrastructure would then step in to provide efficient pricing for these highly specific, potentially illiquid derivatives.”
AI Agents Democratize Financial Tools
Experts believe an intriguing innovation lies in the way AI agents blur the lines between collaboration and competition. While both ultimately benefit the firms deploying them, they also push forward the evolution of data science.
“We may witness the emergence of collaborative AI trading systems that ingest market intelligence data while competing for returns, potentially creating more efficient markets,” Juan Pellicer, senior research analyst at IntoTheBlock, told Cryptonews, adding:
“Advanced natural language processing could provide deeper market sentiment analysis by processing vast amounts of social media, news, governance, and on-chain data in real-time. These AI agents could develop more sophisticated approaches to portfolio rebalancing and risk assessment, particularly in DeFi environments.”
According to Vitomir Jevremovic, founder and CEO of metadata sorting platform Cadastry.io, AI will act as both a passive helper and an active participant in DeFi. It will bolster “security and algorithms for yield farming and risk assessment [as well as] autonomously move value through protocols based on market conditions.”
“AI systems could predict liquidity shifts, dynamically rebalance assets, and execute complex arbitrage strategies, driving maximum value extraction from decentralized financial ecosystems,” Jevremovic told Cryptonews.
Experts agree that AI agents play a democratizing role by making financial tools accessible to a wider audience, beyond just high-end business users.
These agents can handle tasks like signing contracts or completing transactions, allowing users to focus on other priorities.
For general users, agents are a few taps away. With Virtuals Protocol, for example, users just fill out a form specifying the type of agent they need, and it’s ready to work for them.
Top AI agent launchpads and frameworks by number of created agents. Source: DropstabA small amount of cryptocurrency is required to launch such an agent on Uniswap. Some agents—like Terminal of Truths (ToT)—have wallets in their own names rather than running a wallet in trust for human individuals or institutions.
Such wallets are wired into their interactions with humans or other machines, where payment must be received for services given. Experts see AI-enhanced operations as supercharging crypto functionality more effectively than AI cryptocurrencies.
“There is an argument to be made that the most interesting developments will come from non-AI projects that leverage AI effectively, rather than from so-called ‘AI coins,'” said Bezzenberger, who has been building Ethereum Layer 2 (L2) protocols since 2016.
“Many of these AI-driven cryptocurrencies appear to be little more than meme coins with extra steps. The real breakthroughs will likely come from projects that use AI as a tool to deliver meaningful, real-world innovations rather than from projects marketed primarily on their use of AI.”
So ai16z is: a crypto hedge fund run by an AI with $10M AUM … … but none of it deployedSo nothing has actually happened yet, except for the $10M 'capital raise'And it has a $1.3B market cap?
— Mark Jeffrey (@markjeffrey) December 29, 2024The novelty factor, however, will ensure more than passing interest in AI tokens. These include the meme coin, Goatseus Maximus (GOAT), initially poised to be a fad but ultimately shooting off to record numbers.
The Art of the Smart Contract
Pellicer, the senior research analyst at IntoTheBlock, foresees smart contract development improving through AI models.
“Code assistants that will be trained over a vast amount of contracts and technical documentation, which will be able to lower the barrier for development,” he said.
Agents will evolve to be capable of more nuance as testing will be incrementally robust. “Machine learning algorithms are expected to improve blockchain throughput by optimizing transaction routing, block validation, and consensus mechanisms,” Pellicer explained.
Autonomy comes into play when it comes to the art of the deal. The agents can sign smart contracts, fund wallets, or send money on their own. According to Brainbot GmbH’s Bezzenberger, they can also improve trading by simulating threats and scenarios to predict and counter them effectively.
For example, his company has built an encryption protocol called Shutter. The protocol is experimenting with the decentralized artificial intelligence network OLAS to “simulate malicious MEV tactics, such as front-running and sandwich attacks using AI agents.”
“These simulations analyze how transaction placements impact profits or losses, helping us anticipate behaviors, minimize risks, and develop proactive strategies to counter manipulation,” Bezzenberger detailed.
“This approach acts as a dynamic way to stress-test complex economic systems, such as those related to MEV, and could extend to traditional finance use cases as well.”
The Maximum Extractable Value, or MEV, is the maximum amount of value that a group of users can extract from a DeFi protocol.