Hello,

If you told someone in 1995 that in a few decades, they’d be able to order food, book a taxi, or transfer money to friends across the world—all through a device in their pocket, they’d probably be sceptical. Yet here we are, with smartphones having reduced these once-complex tasks into simple taps on a screen.

DeFi stands at a similar inflexion point today. DeFi offers opportunities to earn yields and find new tokens early, but it’s too complex for most people to use. Managing wallets, navigating different blockchain networks, and understanding smart contract interactions can feel like learning a new language. Plus, many hesitate to participate in DeFi because of uncertain regulations. It’s no surprise that DeFi accounts for only 10-20% of total centralised exchange (CEX) spot trading volume. This is because CEXs are much easier to use and have clearer regulations.

This article explores how AI can transform DeFi from a complex ecosystem that serves thousands into an accessible financial platform for millions. We’ll examine how AI-powered interfaces are gradually beginning to bridge the gap between DeFi’s vast opportunities and the average user’s need for simplicity. Though all the DeFAI (DeFi and AI) applications are in their infancy, they offer a glimpse into what DeFi can be: a smooth experience while interacting with financial instruments, from automated trading strategies to conversational interfaces that make complex transactions feel natural.

Let’s start with how financial markets first merged with computers and algorithms. Since the 1980s, algorithms started becoming part of the financial markets in a meaningful way. They’re the building blocks of modern markets. From stock trading to currency exchange.

Algorithms and Financial Markets

Jim Simons comes to mind when I think of algorithms in financial settings. The word “legendary” sits effortlessly before it. He founded Renaissance Technologies, a US-based investment firm that changed the game for quantitative trading. Its flagship fund, The Medallion, delivered an eye-watering 39% compound annual growth rate (CAGR) over 30 years, from 1988 to 2018.

To grasp how extraordinary that is: $100 invested in Medallion would grow to $2.1 million over 30 years, compared to just $1,014 in the S&P 500. The difference is nearly incomprehensible.

But the real magic was in how they did it. Instead of working with Wall Street veterans, the Renaissance Technologies’ team comprised PhDs in mathematics, physics, and other hard sciences. Their approach relied entirely on mathematical models and algorithms to trade the markets—a testament to the power of data-driven decision-making.

This focus on algorithms isn’t limited to hedge funds. Across traditional financial markets, trading is becoming increasingly algorithmic. A recent article highlights that more than 75% of daily currency spot trading, amounting to $5.6 trillion out of $7.5 trillion, is now conducted through algorithms. These systems have reshaped trading desks, shifting the emphasis from human intuition to automated decision-making.

DeFi is in its infancy as far as automation is concerned. In comparison, algorithmic trading has been a part of traditional finance for over three decades. The same data-driven revolution that transformed Wall Street has been knocking at DeFi’s door since 2020.

Algorithms and DeFi

Decentralised exchanges (DEXs) and lending protocols emerged as foundational pillars of this new financial ecosystem in 2020.

DeFi truly came alive when Compound launched its liquidity mining program, sparking an explosion of activity. Around the same time, Aave (then EthLend) saw its TVL and price skyrocket. Several new yield farms launched every day. These farms offered lucrative yields, often paid in the protocol’s native token. But, the value of those yields was tied directly to the token’s market price, adding a layer of complexity to returns. I remember Sam Bankman-Fried’s interview in which he said —

Imagine a magical box that does nothing, yet people throw millions into it because… why not? And as more money piles in, the box becomes valuable—because everyone agrees it is. At some point, sophisticated traders come in and say, ‘Wow, look at all the money in this box! Must be a great box!’ And the cycle continues—until, of course, it doesn’t.

This dynamic created a divide. Savvy traders thrived, navigating between farms, taking profits on tokens, and capitalising on the opportunities. Meanwhile, less experienced participants struggled, often failing to understand the importance of consistent profit-taking in such volatile markets. It became clear that this iteration of DeFi wasn’t designed to scale beyond a small, niche audience.

The need for tools that would simplify DeFi interactions became increasingly urgent as the ecosystem expanded. Lending protocols multiplied, creating demand for aggregators.

Yearn Finance, launched in February 2020, reached a TVL of 2.5 million ETH (around $7 billion at the time). It was a turning point in DeFi’s evolution.

It introduced automated vaults that optimised on-chain yields, offering users defined risk-reward profiles. These vaults allowed users to deposit assets—stablecoins, ETH, and select tokens—while DeFi experts proposed and implemented yield strategies. Funds were then deployed across the DeFi ecosystem based on these strategies, and profits were shared between users, the platform, and strategy creators (essentially acting as fund managers).

This model was a step up for DeFi. For the first time, DeFi felt accessible to a broader audience. Yearn removed much of the manual effort required to participate in the ecosystem while aligning incentives across stakeholders. It was a glimpse of what the next iteration of DeFi could become: efficient, user-friendly, and scalable.

While Yearn made DeFi more accessible, its limitations became apparent as the ecosystem evolved. On-chain yields started normalising, and Yearn’s strategies struggled to maintain an edge. The departure of key innovators like Andre Cronje and tough market conditions in 2022 caused TVL to plummet from its peak to around $250 million.

Yearn was DeFi’s first major attempt at automating yield optimisation, improving manual yield farming by allowing users to delegate capital to experienced managers. But, it still relied on human decision-making. Strategy creators had to constantly track market conditions to identify opportunities, evaluate new protocols, and execute strategies.

It created two major bottlenecks. First, human managers could only process limited market data. Second, scaling to millions of users was impractical due to UX challenges.

AI has the potential to overcome these challenges. By leveraging machine learning and automation, DeFi platforms can now analyse vast amounts of on-chain data, identify patterns, and execute strategies with far greater efficiency than human managers. Using natural language to understand user needs helps increase DeFi’s scale by making it accessible to a large number of users.

Where AI Changes the Game

DeFi provides unmatched optionality, but it remains difficult to use. CEXs are easy to use, but they restrict user control and optionality. AI presents an opportunity to bridge this gap. By automating complex DeFi interactions and streamlining decision-making, AI can make DeFi as accessible as centralised platforms, albeit without sacrificing optionality. On the other hand, AI can help CEXs make quicker listing decisions to provide more options than they currently do.

A practical example of this is Hey Anon, an AI-driven DeFi interface. I tried Hey Anon myself; it was efficient at swapping and bridging, removing the need to find contract addresses or select bridges manually. The entire interaction was chat-based, which makes it accessible to new users. But, it was slower than manually executing these transactions. Also, it currently lacks support for manual transfers—an essential feature that should be incorporated for more flexibility.

Is there a market for DeFi + AI?

Before exploring the intersection of AI and DeFi, let’s take a step back to examine the total addressable market (TAM).

As of Q3 2024, the assets under management (AUM) of actively and passively managed regulated open-ended funds surpassed $80 trillion. By comparison, the combined assets managed by Bitcoin (BTC) and Ethereum (ETH) ETFs stood at $150 billion as of January 21, 2025.

These figures highlight a crucial point: trillions of dollars are professionally managed globally because most people prefer not to handle their finances directly. They gravitate toward products that are easy to use and offer steady growth. Crypto should be no different. We already see this with how user preference is skewed in favour of CEXs.

CEXs still support around five times the volume of DEXs. A significant factor for this disparity is usability. Managing wallets, navigating contract addresses, and understanding on-chain processes are daunting for many. But it comes with massive benefits. Perhaps, the greatest benefit is the possibility of early profit. If you discovered TRUMP under $1 billion on-chain, you were already up five to ten times by the time it launched on a CEX. This is increasingly applicable in the Player vs Player (aka, PvP) phase of the market, where net inflows are stagnant. Assets are exchanging hands among already existing participants.

Rotation is the name of the game. And every week has a new preferred flavour.

Even if you were in crypto for a long long time, it’s highly unlikely that you caught Jailstool or CAR. You had only one day to come to know about it, perform any diligence, buy, and sell—almost always impossible for most humans to execute without prior knowledge. The only way you caught this reliably was by designing a system that reconciles on-chain indicators like newly deployed contracts with volume and price surges and socials like X. Both tokens are down more than 80% from their respective highs and yet to be listed on any major CEX.

One round of price discovery is over. Significant trading activity has already taken place on DEXs and/or OTC desks. Early participants like traders, liquidity providers, or arbitrageurs have already established an informal market price. By the time the asset reaches a CEX, much of the initial volatility and price exploration has already occurred.

Besides, compared to venues like Jupiter and Raydium, most centralised exchanges charge higher swap fees. Jupiter doesn’t charge any fee, while Raydium charges a 0.25% fee on every swap. Moonshot trading app charges users 2.5%, while exchanges like Binance and Coinbase charge variable fees based on users’ volumes. These fees typically range from 0.1% to 0.6%. A pattern emerges from these fees—platforms with better UX can command a premium.

Coinbase has over 110 million users, a far cry from DeFi’s active user base. With this vast gap, the potential TAM for DeFi is vast. If not billions, a conservative estimate suggests that DeFi could aim to onboard a decent chunk of current CEX users if it gets the usability aspect right. This is where AI can play a transformative role.

Enter DeFAI: Simplifying DeFi with AI

DeFAI, an emerging DeFi trend, aims to simplify the DeFi user experience. It will be as simple as talking to the broker to buy and sell stocks—only better. You will interact with an AI agent that can translate text or voice into deterministic on-chain actions and provide you with suggestions that are backed by data.

So, when a token launches on a chain you are not familiar with or have never bridged your assets to, you can go to a chat interface and tell the AI that you want to bridge assets to this new chain to do XYZ action. The AI agent does that for you.

We wrote in our pieces on chain abstraction and smart wallets that both are tools to enhance user experience in crypto. Chain abstraction eliminates the complexities of managing chains and bridges, while smart wallets leverage technologies like passkeys to simplify and secure wallet management.

But, AI agents have the potential to truly expand the DeFi pie. While incremental improvements have been made to address UX challenges, AI agents—if executed correctly—can help DeFi cross the adoption chasm.

Visualising the Impact

Today, the user base of DeFi consists of developers, power users, and late on-chain adopters. With AI agents reducing barriers to entry, the circle of DeFi users can expand significantly, pulling in more CEX users who would otherwise be more than happy to avoid the complexities of decentralised finance.

Abstraction meets intelligence

Abstracting the UX is just one of the things that AI agents can help with. Intelligence is the second. Think about an average CEX user. It is implausible that they already know about on-chain applications they can use and assets they can consider investing in or trading. It has to be curated for them. In the early days of the internet, Yahoo was a curator that helped millions discover and navigate the web. Today’s App Store serves a similar function, determining which apps get visibility and which don’t.

CEXs already act as curators in a way. Choosing which tokens to list and effectively deciding what most retail users can easily trade. If you take this curation away by forcing the user to go on-chain, discovering opportunities and applications will be a daunting task. They need a trusted guide through this complexity. The question is: will AI agents democratise this curation or simply shift the power from centralised exchanges to those who control these agents?

The combination of curation and intelligence is what makes it truly powerful. It’s not enough to simply surface opportunities; users need context, analysis, and execution strategies.

With so much happening on-chain, how does a new user even begin to evaluate opportunities? So many questions need to be answered. Which applications do you use for lending and trading? Where do you buy NFTs? How do you find the correct contract addresses? AI tools/agents like AIXBT can feed into abstraction tools like Wayfinder and Hey Anon.

AIXBT is an agent that devours information on X and puts it in context. It tweets hundreds, even thousands of times per day. Sometimes, its tweets or posts even impact markets. Shlok wrote his thesis on AIXBT. It says that the agent stands out because of how plugged it is into the crypto community, its sophisticated analytical capabilities, and its potential for growth through IP and consumer engagement. The future of AIXBT could see it evolve into a significant player in both the AI and crypto consumer markets, provided it continues to innovate and maintain transparency in its operations.

One of the teams we have been working closely with regards to making retail on-ramping easier is GudTech. Built by a team affiliated with Zircuit, Gud’s vision is to provide contextual information alongside enabling trade execution. Allow me to explain. In the example above for TRUMP token, a user would not be sure that the President of the United States launched a token. Or that multiple large, known whale wallets are buying the token in size. You would have simply seen the ticker on a DEX and purchased it without sufficient context. One of the largest problems in crypto today is that there are 34 million tokens (and counting), but very little context. The crypto sphere is full of unstructured and fragmented data, which is often biased and unreliable.

Gud combines on-chain data with contextual information from social networks to allow directly buying assets on-chain. It solves the problem of reducing the learning curve and cognitive load for new users coming into crypto. You would have been able to see that the asset has rallied 100 times in the last 24 hours and that President Trump had indeed tweeted out the ticker.

In an ideal world, Gud would even verify the contract address and carry out the transaction for you. Gud is building towards an agentic economy where assets from all chains can be purchased, with contextual information from the point of view of a crypto-native user through a conversational interface. The Gud terminal is also capable of critical thinking and able to reason about positive or negative aspects of a trade. In addition, the Gud terminal is free to use for up to 10 queries per day, similar to web2 platforms like Perplexity, and is focused on incentivizing adoption and usage rather than hoarding tokens.

That future may seem somewhat far ahead, but such a model works on two things. First, on how information is captured, contextualised and shared with new entrants to the industry. Imagine having a private wealth manager explaining the latest trends in the industry. This is already happening with sectors like consulting or legal, where spinning up a ChatGPT instance gets you 80% of the insights.

The context needed for such interactions to happen with crypto-native needs does not exist today. Gud aims at bundling it into an easy experience to expand the number of users within crypto today. They are very much a work in progress though. As of writing, the transactional systems for the product are not yet live and the agent on Twitter has had several faulty interactions. But we’ll get there.

Wayfinder is another long-awaited application by the same team that built Parallel, one of the leading games with blockchain rails. Here’s a demo of how the Wayfinder agent aggregated funds from across chains and sent them to a different wallet. Hey Anon has already integrated multiple chains and applications. It combines the capabilities of executing transactions with real time insights from multiple platforms like Twitter, Telegram, and Discord.

Imagine this: you open a sleek interface similar to ChatGPT or Claude, and start a conversation with your personal AI trading agent. You share your risk tolerance, investment goals, and preferences. The agent, understanding your parameters, autonomously manages your portfolio— executing trades, opening positions, and adjusting strategies in real time while staying within your defined boundaries. This isn’t science fiction; it’s the direction we’re heading in. Here is a glimpse of what is possible.

The Reality Check

Applications like WayFinder are not yet accessible to everyone. But before getting swept up in the hype and token prices riding on DeFAI narratives, it’s crucial to take a step back and assess reality. The sobering truth is that we are not there yet. I don’t fully grasp the engineering complexities required to reach our goal, so I can’t predict how long it will take. What is clear, though, is that both intelligence and abstraction in DeFi still have significant gaps to fill.

For instance, take AIXBT, arguably the best intelligence or information synthesising agent in space. It generates several tweets daily, making it impossible to evaluate every investment or trade idea manually. If you had followed all of its calls within the $10 million to $100 million range, you would have achieved an average return of 2%, with a win rate of 39%. This suggests that while AI can process vast amounts of data and surface opportunities, it still lacks the refined judgment of experienced traders. Moreover, this performance comes with an important caveat: a handful of tokens significantly outperform the rest. If you miss those few winners, you are likely to end up with losses from AIXBT’s calls.

Source – SentientMarketCap, shows the performance of AIXBT calls in the week trailing Jan 25.

Given this caveat, it’s easy to dismiss AIXBT’s value. But, this ties into a long-standing debate in traditional finance: does active investing truly outperform passive investing? A Random Walk Down Wall Street popularised the idea that markets are largely efficient and that even professionals struggle to consistently beat an index fund. In fact, studies have shown that monkeys randomly throwing darts at a stock list can generate returns comparable to professional investors. This underscores a broader reality—markets are unpredictable, and human expertise alone does not always guarantee an edge. Yet, The Medallion Fund, which consistently outperformed the market for three decades, proves that human intelligence can create an edge when combined with algorithms.

I personally can’t keep up with AIXBT’s tweets to make trading decisions. However, I would use a screener that distils thousands of AIXBT tweets into the top five trade ideas. Right now, it serves as a decent screener but needs significant optimisation. There needs to be an additional layer that sits on top—one that effectively filters through its output and makes smarter, more strategic decisions. The intelligence challenge isn’t just about volume; it’s about prioritisation. What’s needed is a sophisticated filtering system that refines AIXBT’s numerous suggestions into actionable, high-probability trades.

Where AI Falls Short. At Least, For Now

Taking a step back from intelligence, I wanted to understand how the execution/abstraction side of things works. I tried Orbit to buy a memecoin that it identified as having the highest potential. I interacted with the ‘Meme_Radar_TK_Agent,’ but I didn’t get what I wanted. I had to go back and forth with the agent, clarifying my request repeatedly. Although I picked the AI-suggested token, it failed to retrieve relevant information about the same token. The agent struggled with basic tasks: it would recommend a token but then be unable to pull up critical details about its own suggestion.

Screenshot of my interaction with Orbit

Orbit ($GRIFT) traded at $180 million on January 22. Yet, it couldn’t smoothly execute a straightforward task for a first-time user. This reveals a critical gap between AI’s analytical capabilities and its ability to execute real-world transactions efficiently.

Note — I tried Hey Anon when it was released for the public on Feb 7, 2025.

Of course, the category is still in its infancy, and products will evolve over time. Our own product, SentientMarketCap, is being built in the open, continuously improving based on user feedback and real-world testing.

Similarly, platforms like Griffain and WayFinder may offer enhanced solutions, but they remain largely untested in practical environments. The entire DeFAI space is still an evolving experiment, where products are actively refined through continuous iteration and real-world insights.

What’s clear is that successful DeFAI platforms will need to excel in three key areas:

1. Reliable intelligence that can consistently curate contextualised data to identify profitable opportunities
2. Seamless execution that minimises friction between decision and action
3. User-friendly interfaces that make complex DeFi operations accessible to average users

The technology is progressing rapidly, but we’re still in the early stages of this evolution. The key will be managing expectations while continuing to innovate and improve these systems based on real-world performance and user feedback.

The application of AI in DeFi is not without risks. Poorly trained models, reliance on historical market conditions, and the potential for manipulation are all concerns that need to be addressed before AI-driven DeFi platforms reach mass adoption.

Learning from Feynman

Richard Feynman’s argument on machine intelligence is highly relevant to DeFAI. He argues that a machine can be better than humans at specific tasks. If we can combine these specific tasks into a superset—a new system—it can significantly aid our decision-making and execution in financial markets. AI in DeFi should follow this principle: rather than replacing human intuition, it should enhance our capabilities by integrating multiple intelligence layers—automated execution, market analysis, and risk assessment—to create a seamless experience for users.

This modular approach to AI capabilities has deep implications for DeFi’s evolution. DeFi doesn’t need just automation—it needs intelligence that optimises execution. Take an example of a well-run hedge fund. It has different teams with expertise in specific areas. Some focus on executing trades with minimal slippage, others analyse patterns to predict market movements, and a third team ensures capital flows efficiently across different markets.

AI agents in DeFi can function the same way. One agent could specialise in executing trades efficiently by reducing price impact and avoiding MEV attacks. Another could detect patterns in on-chain data to anticipate liquidity shifts or market trends. For example, this agent can be plugged into tools like GMGN and Cielo, where it can track wallets on-chain to aid its other analyses. A third could manage cross-chain transfers to ensure funds are optimally allocated across ecosystems. When combined, these agents go beyond plain automation. They bring intelligence to execution— from providing inputs into what to trade to ensuring trades happen at the best possible prices, with minimal risk, and across multiple networks, seamlessly.

Towards Agent Collaboration

Most DeFAI products are attempting to tackle both intelligence (analysis, synthesis) and abstraction (execution) capabilities, and for good reasons. Either component alone provides limited value, much like having a map without a vehicle or vice versa. But, the real power lies in specialisation and integration.

The current landscape resembles a fragmented ecosystem where different agents excel in distinct areas. Some demonstrate exceptional skill in market analysis and pattern recognition, while others excel at executing complex DeFi transactions. The optimal solution likely involves agents working together and leveraging each other’s strengths. Imagine Anon’s expertise in DeFi integrations combined with AIXBT’s analytical capabilities— this collaboration could create a seamless experience where market insights smoothly translate into executed trades.

Listen is building in this direction. The idea is to create a system where multiple AI agents with specialised functions collaborate to manage the intricacies of DeFi. By integrating these agents, it aims to automate not just individual tasks but end-to-end financial strategies.

This approach would allow users to issue complex commands like portfolio rebalancing or yield farming across multiple protocols through a simple conversational interface (voice and text), making what was once a daunting task for even seasoned DeFi users, accessible and manageable for the average person. The partnership with Arc is aimed at enhancing capabilities by providing a platform where these AI agents can interact, learn, and scale. This ensures that the execution and intelligence layers are not just separate but work in concert to provide a holistic DeFi experience.

A Familiar Evolution

The current state of DeFAI is reminiscent of the early days of banking. Initially, financial services were fragmented—users had to visit separate institutions for bill payments, investments, and transfers. As banks came online, integrated platforms emerged, offering seamless financial management in one place.

What DeFAI needs is its own “super-app” moment—platforms that seamlessly integrate various specialised agents. Think of it as an orchestrated system where analysis agents provide market intelligence, execution agents handle transactions, risk management agents monitor positions, and portfolio optimisation agents balance allocations.

This integration would create a unified experience where users interact with one interface while multiple specialised agents work together behind the scenes, much like how modern food delivery apps handle everything from restaurant discovery to payment processing. The future of DeFAI is about creating ways for specialised agents to work together smoothly. This approach would allow each agent to focus on its core strengths while participating in a larger, more capable ecosystem.

Lowering Barriers, Unlocking Adoption

Robinhood revolutionised retail investing by making stock trading accessible to millions who had never considered participating in the markets before. COVID struck, and in the first four months of 2020 alone, Robinhood added more than 3 million new funded accounts. 1.5 million of those were first-time investors. This unprecedented growth was driven not just by commission-free trading and an intuitive mobile-first design but also by external factors like stay-at-home orders during the pandemic.

DeFAI has a similar opportunity. The complexities of DeFi have long been a major hurdle for widespread adoption. Cumbersome wallet setups, confusing interfaces, and fragmented liquidity across multiple chains discourage all but the most dedicated users. If DeFAI is to thrive, it must follow Robinhood’s playbook—removing friction and making DeFi as simple as opening an app, selecting an asset, and executing trades in seconds.

Beyond usability, AI-driven curation will likely redefine discovery within DeFi.Just as Yahoo once curated the early web and app stores guide mobile discovery today, I am curious about how new business models emerge around AI-powered DeFi curation. The open question is whether these innovations will empower users or simply shift control from centralised exchanges to those who build and manage these AI systems.

We are still in the early innings of AI in DeFi. The coming years will determine whether these technologies truly democratise access to decentralised finance or, paradoxically, introduce a new form of gatekeeping. The challenge isn’t just about automation—it’s about ensuring that AI enhances accessibility, transparency, and decentralisation rather than replacing one set of gatekeepers with another.

Waiting to use new age DeFAI,
Saurabh Deshpande

Disclaimer:

1. This article is reprinted from [Decentralised.co]. All copyrights belong to the original author [Saurabh Deshpande]. If there are objections to this reprint, please contact the Gate Learn team, and they will handle it promptly.
2. Liability Disclaimer: The views and opinions expressed in this article are solely those of the author and do not constitute any investment advice.
3. The Gate Learn team does translations of the article into other languages. Copying, distributing, or plagiarizing the translated articles is prohibited unless mentioned.