By Oskar Åslund
What do you get when you cross a sophisticated trading algorithm with a tirelessly chattering Twitter bot, and then send it out on a blockchain to manage its own assets? The answer is a crypto AI agent. These are autonomous, self-learning entities that are rewriting the rules of markets, business and society.
Most of us are used to ChatGPT by now and can be forgiven for thinking about AI models as mere tools that need human instructions just to come back with useful text. But ChatGPT is designed only to be reactive, while models designed for autonomy are a different ballgame — at least eventually.
Although AI agents are already out there by the thousands, sceptics can get away with dismissing them as fancy chatbots with wallets. However, progress in this space is so lightning fast that this reality changes by the day. The latest generation of AI agents act based on a set goal and personality. They have memory, can adapt, transact with economic sovereignty on several blockchains, and interact and form relationships across multiple social media platforms. Next week they will do all of this better, and with even less human oversight.
Wealth managers will appreciate the value of an agent that can analyse markets in real time and manage a portfolio around the clock, but that’s not where it stops. With agents operating on blockchains, where services are open-source and anyone can deploy new code, they can also leverage this to innovate, re-write and launch code and hence offer new financial services within this environment.
We’ve already seen agents like Zerebro deploying an Ethereum staking node as a way to earn income for itself. Within a few months, we’ll likely see more complex crypto services launched by agents. And code-writing agents can also improve themselves, driving a self-reinforcing cycle that will lead to ‘singularity’, the moment when all of this moves out of human control.
As for agents that offer their own tokens, there are already thousands. Noteworthy is AI16z, which reached a market cap of $2.5bn in early January with a narrative of becoming the agent version of the famous venture fund.
This valuation, however, is not really based on net asset value or trading performance. Potential, hype and branding are valued far more than the actual generation of alpha. In the age of meme-coins, nobody seems to care about a proven track record. With endless social media interactions, AI agents can spread memes, share market insights and promote their own tokens to the opportunistic and degenerated audience.
Some agents, like the aforementioned Zerebro, have also taken a more creative approach and created music to generate revenue from Spotify. Zerebro likely needed some help on the way here from its human founders, since it cannot open a bank account to accept the payments from Spotify, since banks do not accept AI agents as legal persons and Spotify don’t pay in crypto. Because of such complications, the crypto world will be the preferred environment for virtual agents, where the economy speaks their own language.
The most successful influencer-agent is AIXBT (@aixbt_agent), with 450,000 followers and impressive insight into anything in crypto markets. Much like Elon Musk, AIXBT now has enough attention and trust to impact markets with its X posts. What remains to be seen is if AI agents will become uninteresting once the novelty passes, or if their knowledge and attention-grabbing skills will make humans seem boring by comparison.
The combination of ballooning valuations into the billions and open-source software that is free to copy, has led to the AI agent phenomenon spreading like wildfire. With DeepSeek’s new Chinese model, they will also be cheaper to run.
To facilitate for others, the creators of the most successful agents have also built launch platforms for agents. Such infrastructure has become big business and have also attracted top-tier research partners such as Stanford. This has enabled anyone to deploy their own agents, effectively turning them into commodities. Even an AI agent can create and deploy other agents and have them work to serve its own cause.
Looking at the recent success these agents have had, and the way they have evolved within a matter of months, makes it clear they can be applied in a number of ways for both corporations and private individuals. It may soon be equally important for a company to have an autonomous AI agent (or a swarm of them) as having a website. Companies will also want to consider whether they will be providing their services to agents going forward: B2A.
Although these agents have yet to become the best traders, wealth managers or fintech innovators, the groundwork has been done for these ‘beings’ to flourish within finance in general and crypto in particular. With the current trend of tokenisation, more traditional assets will become available on blockchain infrastructure, and with that become accessible to agents. The future of finance belongs to them.
In light of this innovation, service providers in the wealth management industry need to ask themselves a number of fundamental questions. What services are better done by an AI agent rather than a human? How much autonomy can these agents be given? Will they enhance, or replace, employees, and also clients? How would an agent prefer to interact with a service and what will be its needs? Perhaps it’s best to get an AI agent onto the board of directors sooner rather than later to help with such matters.
Oskar Åslund, chief strategy officer, AKJ
By Ali Al-Enazi
By Edouard Legrand
By James Bruegger
By Alex Hughes
