AI Agents Will Drive the Next Wave of Crypto Adoption

It’s likely that the online financial tools we use today will seem outdated in the future, as more and more people manage their finances through intelligent bots or agents that handle transactions for them.

Luke Saunders is a Partner at Delphi Ventures and CTO at Delphi Labs.

It might seem implausible, but agents, including liquidation bots and arbitrage bots, are currently in operation. These agents function autonomously, executing tasks such as liquidating positions or capitalizing on price differences between markets for their creators’ benefit. Though they cater to specific niches, there exist thousands of these entities actively engaging in on-chain activities on behalf of others.

Bots and agents lack the human ability to appreciate aesthetics or interact with apps through clickable buttons. Traditional finance (TradFi) offers APIs, but these tools can be underdeveloped, shrouded in secrecy, and come with limited instructions and inconsistent implementations. Navigating their use is a complex challenge for both bots and humans alike.

In my latest exploration of X, I highlighted the challenge that crypto user experience poses for humans. However, from an agent’s perspective, the inherent transparency, open-source, and programmable aspects of crypto are invaluable. Here, code holds a privileged position. Agents have direct access to both the source code and essential data, which is not typically the case with TradFi APIs that remain half-baked. The most advanced agents will be developed on the most robust platforms, and I firmly believe that crypto offers the best foundation for such platforms.

How we get there

Although we haven’t reached that stage in reality, there is a viable route leading us there. Let’s retrace our steps and focus on the current agents before delving into the means of transitioning towards that state.

These previously mentioned software entities, such as liquidation bots, can be categorized as agents due to their autonomous nature. However, they are rather fundamental in comparison to more complex agents. They are typically designed for a singular, restricted function and lack the sophisticated interface of a ChatGPT model for receiving instructions and providing feedback.

As an analyst, I would rephrase it as follows: I’ve observed that large financial firms such as Jump and GSR possess the resources to construct intricate systems capable of autonomously managing funds and addressing certain issues. However, the practicality of this approach for smaller entities is limited due to their budget constraints and fewer development team members.

Technology continually advances to make complex tasks more accessible. What once required a large computer and a team of experts can now be accomplished by an individual using just a smartphone.

With the increasing use of AI, the intricacies of deciding which smart contract functions to invoke and interpreting data within a contract’s current state can be automated. Consequently, creators and owners are granted the freedom to provide more elevated commands, while the agent takes care of executing them in detail.

Today, language models like GPT-4 demonstrate the ability to grasp the meaning behind user queries and generate intelligent responses, while also accessing vast amounts of internet knowledge for context. Therefore, envisioning a textual interface for an agent that not only processes crypto-related media but also absorbs data from all blockchains and carries out on-chain actions is a logical progression.

Think about what people in crypto are spending all their time doing today:

An agent can function as an adept airdrop tracker, dedicating their entire day to identifying budding projects with potential airdrops, speculating on profitable actions, and executing those steps accordingly.

As a meme analysis expert, I follow a specific algorithm to discover emerging meme coins that exhibit traits of a devoted community. Once identified, I proceed to purchase these assets. Subsequently, I monitor the market trends and sell when it appears the level of interest is starting to decline.

Certainly, Above all, reflect on the fundamental ways that the latest AI agent interface streamlines finance essentials:

As a researcher, I would recommend exploring money market funds or high-yield savings accounts that offer competitive returns with minimal risk and quick access to your funds. These options typically provide yields higher than traditional savings accounts without the long-term commitment often associated with certificates of deposit (CDs). Since you require instant access to your USDC, avoiding CDs is essential. Let me help you find some viable alternatives that fit your criteria.

As your dedicated analyst, I suggest allocating an initial investment of 50% towards Aave, 20% each to Ethna and X, and 10% to Y. I will closely monitor market conditions, keeping a keen eye on their respective risk levels and making adjustments to the distribution as needed through rebalancing.

You: “Sounds good, do it.”

As a researcher exploring the future of blockchain technology and digital wallets, I ponder over the question of how MetaMask or centralized frontends can be dethroned. My hypothesis is that in the not-so-distant future, we may no longer require traditional user interfaces as our in-agent wallets take center stage.

In the financial world, Traditional Finance (TradFi) intermediaries will persist, but those established on cryptocurrencies will surpass them in terms of user experience. This distinction is crucial because as I anticipate an increase in the usage of agents, the superior offerings from crypto-based agents could attract a larger number of users, transactions, and value to decentralized platforms.

As a crypto investor, I’ve often pondered over the notion that cryptocurrencies seem to be in search of a real-world application, but perhaps they are perfectly suited for powering the financial systems of the future – those run by machines.

Here’s a suggestion for paraphrasing the given text while maintaining its original meaning: