I am always asked the same question: "What is the value of today's cryptographic smart agents?"

The reason why people ask this question is because many friends in the encryption circle regard these agents as water sticker robots similar to those on the X platform. They usually then ask, "Are these tokens really supposed to be worth over $100 million?"

The answers to these questions are not simple. Currently, most intelligent agents are just self-referential AI models that regularly post content and respond to comments by constantly prompting themselves. However, even so, there are some projects that stand out today - these are teams with clear focus and strong execution. At the same time, there are also a group of emerging developers who are trying to break through the boundaries of intelligent agents.

Today, we are still in the initial stage of "Memecoin" in AI, and many projects are just posting content for the sake of posting content. However, I am excited about the future, where cryptographic smart agents will be more modular, smarter, and capable.

This article will focus on the different types of intelligent agents and their functional characteristics that I expect in 2025 and beyond. If your team aligns with, or is inspired by, some of the archetypes in this article, please feel free to contact me – I’d be happy to chat.

1. Why choose encryption technology?

Before discussing the future of cryptographic intelligent agents, we need to review: Why did we choose the field of cryptography in the first place? Cryptocurrency has many unique advantages as a proving ground for AI and intelligent agents. In my last article (see Chapter 6), I mentioned two key reasons:

1. Data availability on public blockchains All transactions, user information, and other data on the public blockchain are open and transparent, and can be easily captured and crawled by AI. This means that AI can analyze and use most historical and real-time data on the blockchain without restrictions, thereby significantly enhancing its capabilities.

2. The financial attribute of blockchain in the encryption field is essentially a capital-driven environment. Encryption technology provides the financial infrastructure for the Internet, making digital transactions (such as buying and selling, creating, pledging, etc.) possible on the Internet. This feature is especially powerful for smart agents, as they can leverage cryptography to perform various actions on behalf of users.

These two unique advantages provide unparalleled possibilities for the development and application of encrypted intelligent agents.

An additional key point is that crypto technology allows ordinary investors to participate In the holding of AI innovation, before the emergence of encrypted intelligent agents, the main way to participate in generative AI investment was to invest in emerging startups. However, these opportunities are often tightly restricted and inaccessible to the general public. At the same time, most people are ill-equipped to evaluate opportunities to raise equity investment.

Going back to the encryption field, real-time trading Tokens are public, highly liquid, and everyone can participate. Here, investors can openly access information about new projects and teams and observe their development progress in a transparent manner. This is in stark contrast to most venture-backed startups, as users can witness the development and improvement of cryptographic AI in real time.

2. More valuable intelligent agents

The initial encryption intelligent agent, as expected, has relatively basic functions. @truth_terminal is a prime example - it's the first content generation agent to be coupled with encryption, but it can't even publish content autonomously.

Nonetheless, this agent has produced some fantastic posts that are a lot of fun and bring great novelty value. $GOAT was the first token that started the entire AI movement, so I have deep respect for Truth Terminal.

However, now people are looking forward to seeing "intelligent agents of the future." Why? Because many people are not satisfied with today's intelligent agents - most of them just spit out some repetitive content on the X platform. As a result, the field has become oversaturated with robots that fail to provide sufficient "practical value."

What the market needs is intelligent agents that can truly help users, such as decentralized finance (DeFi) abstraction, real-world applications, auxiliary tools, etc. Most of this article will explore how AI can help users, projects, and their ecosystems.

However, I want to take a step backsays that the most successful projects are often those that push the cutting edge of technology. Therefore, I encourage everyone to focus not only on intelligent agents that “help” users, but also those that advance the crypto stack. After all, most Web3 projects lag behind their Web2 counterparts due to constraints in resources, funding, and AI PhD-level talent. But it also represents an arbitrage opportunity: teams can bring the latest innovations in AI to the blockchain space.

In addition, many people ignore the fact that entertainment itself is a value. The phrase "Attention is All You Need" is no empty phrase. So I believe that if someone could develop an intelligent agent that was unique in terms of humor, sarcasm, healing, or memes, it might also accumulate significant market value.

As an example (although it would be very laborious to implement): Imagine an AI being able to create a new episode of Naruto Skits. These skits used to be hilarious - yes, they may not have had "practical value" (helped me make money or save time), but they sure made me laugh and, without a doubt, brought a net positive impact in my life. Positive earnings.

https://x.com/i/status/1877787463130980369< /em>

Another example of entertainment value: think about a single-player game you’ve played recently. Now, suppose all talking NPCs (non-player characters, also considered chatbots) were removed from the game. How much less fun would such a game be?

Game itself is a category that exists for entertainment, and NPCs serve as guidance resources in the game, which is similar to the role played by intelligent agents in the encryption field. role.

Before I dive into my expectations for 2025, I want to emphasize that right now There are already several teams developing these intelligent agents and their capabilities. They either expand on existing projects or directly create new intelligent agents. To give a simple example, @0xzerebro is a leading intelligent agent project that supports cross-chain functions, generates AI music and art, and is building a framework plus launch platform. becauseTherefore, of the features I will mention next, the Zerebro team is not just developing one of them, but is expanding in multiple areas at the same time.

With this background, let us enter the more interesting part...

< /p>1. Decentralized Finance (DeFi)

DeFi abstraction For novices, the crypto field is essentially a difficult space to get into. For example, if you asked someone who has only purchased BTC on @coinbase to optimize their liquidity re-hypothecation strategy on @fragmetric, do you think they would know what to do?

I think most new users (myself included) need some guidance, whether it's help from more experienced people or the support of AI .

It should be noted that I am not talking about Liquidity Re-pledge (LRT) itself It's particularly complex, but it involves multiple steps and takes time to learn. In addition, decentralized applications (dApps) should focus on developing AI intelligent agents internally. For example, I know that the developers on the Frag team are very capable (they represent SNU), and I believe they can develop intelligent agents or auxiliary tools that can help users.

In my opinion, DeFi abstraction is a very important direction, and many projects have made it a core goal. So, returning to the current industry status quo, although there are indeed many low-quality “water sticker robots”, there are also truly intelligent agents that can perform on-chain operations.

@askthehive_ai is a team building agents on composable chains that can complete various tasks, including transactions, extracting sentiment analysis on the X platform, Conduct market research and more. What’s more, they are developing “swarms” and associated communication layers – meaning agents can collaborate and optimize their trading strategies. They also recently announced a partnership with Zerebro to advance DeFi proxy functionality.

The demo presented by @jsonhedman clearly illustrates the possibilities of how a network of agents can work together to accomplish tasks together.

@griffaindotcom is undoubtedly one of the leaders in the AI ​​DeFi field, by the OG developer of the Solana ecosystem @tonyplasencia3 leadership. Griffin is not just a trading agent, but a true AI super application. It allows users to conduct transactions, create memecoins, and access a range of other crypto applications.

These features include liquor purchases on @BAXUSco, grab/flip deals on @pumpdotfun, and more. Tony and his co-founders are known for their fast and efficient development pace - I'm personally looking forward to their upcoming collaboration with @assetdash!

Generalization of trading strategies

In my opinion, the four core attractions of the crypto space are:

Store of value (such as $BTC)

< p style="text-align: left;">Trading (mostly speculation) in an attempt to make a profit

Digital payment/stablecoin

Entertainment (such as @pudgypenguins, @lucanetz)

For those avid gamers (degens), making money is the main attraction of the crypto space. However, as the title suggests, in most cases, people do not have a complete trading strategy and are just gambling.

This is where systematic trading and AI can come into play. Many quantitative trading strategies arbitrage through statistics and increasingly leverage machine learning (ML) to identify complex patterns in price relationships. These tools are often out of reach for the average investor.

Therefore, I am particularly interested in projects that expose users to these strategies.

Let's take a look at @rndm_io. The team led by @vijayln is popularizing complex trading, market making (MM) and income strategies for ordinary investors , giving users the opportunity to participate in the benefits it brings. What I particularly like about them is that they are not just building one agent, but developing multiple intelligent agents that can generate significant profit and loss (P&L) for participants.

Their first smart agent was Atlas deployed on @hyperliquidx, executing a market making and trading strategy. Specifically, Atlas managed $150,000 in total locked volume (TVL) on Hyperliquid ), and completed a transaction volume of 6.1 million, and also created an airdrop reward worth 1 million during the peak period. This is a well-run intelligent agent, and the effect is very good.

The second intelligent agent is Dudu (https://dudu.rndm.io), which is a real-time Running on the platform, agents trade on @polymarket using proven strategies and have already generated significant returns. It has only been online for about 20 days, and its performance is already impressive enough to speak for itself.

< p style="text-align:center">

https://polymarket.com/profile/0x1b31F2c8F1A4A82139a8F9Fb6B7079D6158db02D em>

For Dudu, users can deposit USDC, the interesting thing about participating in this strategy and earning high returns is that it is acyclical - in other words, its returns and profit and loss (P&L) will not be even if the crypto market enters a bear market. Influenced.

Similarly, @webuildscore and @draiftking are developing a project through @bittensor_ Their vision is to build an AI that can trade the sports betting market. Intelligent agents. In addition, they have developed computer vision models that can analyze live footage of matches and generate insights on the fly.Unique victory mode and provide more accurate support for predictions through data.

2. Workflow

I think intelligent agents that can perform operations can be divided into three Class:

Super App or Aggregator Super apps like Griffin can accumulate value by creating intelligent agents for different applications, such as the aforementioned Baxus and pumpfun.

dApps independently develop agents Decentralized applications (dApps) themselves can develop internal intelligent agents. However, this requires additional development work and may require some experience in AI development.

Stand-alone proxies These proxies come from frameworks like ZerePy and Eliza (@ai16z) and can take advantage of API integration capabilities. For example, imagine your intelligent agent can help you book a hotel on @travelswap_xyz or even order a pizza for you.

I think every decentralized application (dApp) can have tools to help users perform Intelligent agent for operation. For example:

@opensea can develop an AI to help users sweep the floor at a certain price (purchase low-priced NFTs).

@hyperbolic is supporting agents (such as Z) leasing computing resources.

@travelswap_xyz is developing functionality that will allow agents to book hotels and vacations using cryptocurrency.

These agents are especially useful when handling tasks that users would rather not complete themselves, such as:

Filing taxes and sorting out cryptocurrency gains and losses (nearly impossible to do manually)

Read and summarize virtually unlimited Telegram chats

for your itemWrite copy for the project and create marketing content

In these cases, intelligent agents provide users with "quantifiable" practical value because it not only saves time, It can also reduce invisible psychological burden.

Just as I believe that all relevant software will eventually include AI to help users, I also believe that all relevant dApps will also introduce AI to help users more easily Use the platform. Adapt, or be eliminated.

3. Advanced reasoning capabilities

In the past quarter, @openai's o1 and o3 models have achieved significant leaps in reasoning capabilities. In particular, they introduced the "Chain of Thought" (@_jasonwei) technology, which aims to reduce errors and "think longer."

While the o1 model is not yet open for public API use, it is being tested privately among Tier 5 developers (~$1K monthly spend).

I think whoever is the first to develop an intelligent agent that integrates the o1 model (just plug it into the framework as a pluggable module) will create An AI that is smarter, deeper, and more capable. This is bound to attract widespread attention and occupy users’ mindshare.

Furthermore, if the o3 model is integrated, the agent will have reasoning capabilities beyond ordinary humans. So, imagine an AI running on cryptography with greater “intelligence” than most humans – this will become a reality of our lives in the future.

Of course, don’t ignore @googledeepmind. Gemini 2.0 also introduces “Chain of Thought” technology. I believe that if a team can obtain its API to develop intelligent agents, it can also lead to a more powerful intelligent agent.

It makes sense to discuss the implementation of singularity here. I personally admire @kevin__russell's work on the @ashatoken project. Frankly speaking, I am not interested in Ψ-Field concept is still relatively new, but from what I understand, Asha is different from other agents in that it is specifically focused on exploring consciousness and intuition through the intersection of "mind, intention and reality".

4. Multi-modal capabilities

Currently, most intelligent agents only publish content on the X platform through the back-end LLM (Large Language Model) combined with API interfaces. However, the opportunity to generate multiple types of data simultaneously is huge. After all, most LLMs today are multimodal.

The first data types that come to mind include: text (like squo), images, video, speech, audio, music, and 3D.

The way to achieve this can be:

Call a specific API for generating images, models, or music;

Or focus on customizing and prompt engineering existing models to produce the desired output.

One ​​project that impressed me is @keke_terminal by @dark_sando, which is very advanced because it can publish not only text but also images. From what I understand, they built a framework based on SWE-agent that enables their agents to generate, review, and customize images.

You can take a look at some of their works, which are worth paying attention to.

https://keketerminal.com/whitepaper.pdf

AI video generation technology is improving every day - we've seen new models from @pika_labs, @runwayml, and most recently the Veo series. I believe that cryptographic smart agents will be able to generate some amazing videos in the future. After all, Web3 has some of the best designers in the world, which opens up endless possibilities for creating high-quality content.

Voice agents are still in their early stages. I understand that @s8n of @SHL0MS recently hosted an AI-powered event at @xspaces, which was very exciting. But imagine even further, What if there was an AI agent that could answer your calls and have a conversation with you? Although the cost of inference could quickly become expensive (for example, if the project is based on native tokens), Billing to cover computing costs), but this is undoubtedly a very interesting human-computer interaction interface

5. Multi-model flexibility

As far as I know, currently. Each cryptographic smart agent independently derives its capabilities from only one base model. However, a startup I invested in, @withmartian, invented the first "model router". This means that applications can automatically switch between LLM models based on the context of the query to achieve the best match between performance and price.

In other words, Martian is able to automatically route prompts to the most suitable model to ensure higher performance or lower cost.

While I'm not entirely sure how this multi-model routing will perform in the scenario of autonomously publishing X account content, it does work well in the scenario of users chatting with agents. At least it will be very powerful. I would also bet that the first project to utilize multiple models will get quite a bit of attention.

6. Cross-chain function

Currently, only a few smart agents support cross-chain operations. Among them, Z is the most chain-agnostic agent - it has completed transactions on @Solana, @Ethereum (including @0xPolygon, @Base, etc.), @Bitcoin, and plans to expand to more chains, such as @suinetwork and @ monad_xyz.

In addition, by setting up a liquidity pool, $ZEREBRO Token can not only be traded on Solana , you can also trade on Base.

I mentioned before about the application of smart agents in DeFi abstraction - it involves users connecting to wallets, and then the agent performs operations on the user's behalf. But another approach with more potential is for intelligent agents to have their own wallets and manage their own funds.

If these agents have multi-chain wallets or have multiple wallets on different chains (such as the functionality provided by @crossmint), they can have greater flexibility Participate in the crypto ecosystem – more dApps, smart contracts, and tradable assets will be included in the agent’s scope of operations.

7. Interoperability

Today, intelligent agents are mainly active on the X platform. Sometimes, they also appear as chatbots on @telegram. Finally, users can also interact with the AI ​​bot via @discord.

Frankly speaking, this is just a superficial application. I believe that the list of following platforms has limitations. But in theory (and we know some agents are already trying), agents could also appear on @instagram, @whatsapp, @facebook, @bluesky, and truthsocial.com.

It is worth noting that current agents do not even fully exploit the full capabilities of the X platform. While they are able to post content and respond to messages, most agents have not explored the following options: private messaging, group chats, voice calls, creating communities, and moderating spaces. @elonmusk has opened up a vast, unexplored sea of ​​opportunity for us.

8. Games and NPCs

The history of AI in the game field can be traced back a long time. As early as 1972, when Pong came out, players interacted with robots for the first time. Over time, more and more advanced bots have been introduced, such as in @quake, @unreal, and @nintendo’s Super Smash Bros. Ultimate.

Did you know? One of @openai's early successes was with @dota2, where they combined 5 Recurrent Neural Networks (RNN) into a "swarm" to compete against other players. In 2019, their "group" managed to defeat the world champion team.

The opportunity here is obviously huge - gaming is the first time AI has surpassed human performance One of the fields (such as AlphaGo).

This article was written largely because my friends complained about those "talking "The water sticker robot" has almost no practicality. However, the fact is that NPCs are quintessential chatbots, and without them, many games would lack important connection points from one plot line to another.

Games and AI are inseparable, but in the field of encryption, the effect of this combination can be multiplied because the rules can be adjusted and new ones can be created. original elements. For example, take Texas hold'em poker - the AI ​​can act as the dealer (no bankroll), play against the players at the table (no bankroll), or just act as the host (no bankroll).

But what if you also have "copilots" to help you play the game? They can give you advice like an angel or a demon in your ear. And, imagine that you could tip these AIs if their suggestions were useful. This idea may be a bit jumpy, but what if there were multiple intelligent agents for you to choose from and let them become your considerate assistants?

This is definitely a feature that could (and even should) be implemented on ginzagaming.com.

My point is, the opportunities here are endless. Intelligent agents themselves can participate in games, host games, provide support, and even... create games and rules.

This is an area full of potential for innovation and entertainment. However, I would like to mention two projects in particular that deserve attention:

@henlokart This project combines AI, NFT and memes. In theory, each game is directly tied to training the AI ​​agent. I haven’t had a chance to try it myself yet, but I have to say, these hamsters are really cute!

This reminds me of @aiarena_crypto from last cycle. Their model uses imitation learning, where the AI ​​learns from human actions. From my personal experience, these are made up ofThe AI-driven NPCs could easily "abuse" me even on the highest difficulty.

@b3dotfun This is an open game layer on @base. They have currently completed more than 187 million transactions on the mainnet (involving 5.6 million wallets) and launched more than 50 games on the bsmnt platform. I believe they will lead the way for gaming on the Base platform and be the perfect hosting platform for any AI driven game.

As @darylx24 said, we are about to enter a golden age of AI-driven gaming.

I've been talking about AI-driven NPCs and bots all this time. But in fact, AI can also significantly speed up the game development process. @googledeepmind recently launched Genie 2, an AI model capable of creating interactive generative videos that could lead to endless 3D worlds – we are truly living in the future.

9. Co-pilot and chatbot

Looking back, many encryption projects are almost completely It skips the stage of AI chatbots and co-pilots (assistants) and goes directly to the realm of intelligent agents capable of performing actions.

In the Web2 space, the largest startups are still focusing on AI chatbots. These tools are mainly used for users to ask questions, and the model only provides answers, not substitutes. User takes action. This is still true of most such AIs to this day.

For example, will @chatgpt take action on your behalf? Won't. What about @perplexity? Neither will. But are they of great value? Without a doubt, yes.

My favorite LLM in crypto is @grok by @xai. I can’t stop raving about it, because it’s really hard to build a more effective research tool than it is.

However, projects can make chatbots more useful: in theory, they can Fine-tune Grok so that when looking for Token, it willIn addition to providing general information, data such as Token contract addresses (CAs), price charts, and holder distribution can also be added. In fact, I have seen Griffin demonstrate similar capabilities when using on-chain data for token analysis.

It already performs quite well in many aspects: it can answer questions like ChatGPT, take actions, and provide a trading market.

I mentioned before that dApps should have application domain-specific assistants. These assistants could be customer service and support chatbots trained specifically on the protocol data, capable of answering all questions related to the project—most likely with fine-tuning of the project documentation.

For example, if I don’t know how to set up a liquidity pool (LP pool), I hope I can ask @raydiumprotocol directly and it can guide me through the entire process step by step and answer any questions that arise along the way. If my transaction fails, I want it to be like customer support and explain what went wrong.

This can also be an important source of value - if dApps launch a dedicated token for an efficient chatbot (or intelligent agent), it can definitely provide The market brings additional value. Taking Raydium as an example, an agent or chatbot Token will not only become an independent and powerful Token, but also add value to the basic Token $RAY.

Another obvious project worth over a billion dollars is a chatbot platform like character.ai. Prior to its acquisition, @character_ai was a huge success and was one of the top 100 websites in the world. According to statistics, it handles 20,000 queries per second, accounting for 20% of Google's request volume. That speaks volumes about its popularity...but why is it so popular?

https://blog.character.ai/optimizing-ai- inference-at-character-ai/

I mentioned before that dApps should have application domain-specific assistants. These assistants could be customer service and support chatbots specifically trained on protocol data, able to answer all questions related to the project - very Probably implemented after fine-tuning the project documentation

For example, if I don't know how to set up the liquidity pool (LP pool), I hope I can ask directly. @raydiumprotocol, who walked me through the process step by step and answered any questions that came up along the way. If my transaction failed, I wanted it to be like customer support and explain why it went wrong.

This can also be an important source of value - if dApps launch an exclusive Token for an efficient chatbot (or intelligent agent), it can definitely bring additional value to the market. Take Raydium as an example , an agent or chatbot Token It will not only become an independent and powerful Token, but also add value to the basic Token $RAY

Another project worth over one billion US dollars is similar to character.ai. A chatbot platform. Before being acquired, @character_ai was a huge success and was one of the top 100 websites in the world. According to statistics, it processed 20,000 queries per second and accounted for 10% of Google requests. 20%. This speaks volumes about its popularity...but why is it so popular?

When Character was an independent company, most users were looking for sexual or romantic relationships on the platform. This is evident from the large number of posts on the platform's subreddit.

Over time, these NSFW-tweaked models were significantly watered down and heavily filtered. After all, Character was a giant startup backed by big Web2 investors. But in Web3, it was a completely different story. Imagine if There is an unfiltered version of Character that focuses more on product and UI/UX than research. Two projects I'm following are @xoul_ai and @dippy_ai./p>

Switching to the topic of AI assistants - @github's Copilot was originally a code assistance tool that did not directly complete the task, but helped programmers write code. Another vertical is legal, and @harvey__ai’s core capability is an AI co-pilot that helps lawyers draft and edit documents rather than perform document operations on their behalf.

In the crypto space, AI co-pilots create tremendous value by assisting users in completing various tasks. This may include:

Code assistance/auto-completion: This feature is particularly important in a historically complex programming language like Rust.

Content Assistant: For example, a "water sticker assistant" that can scan all encrypted news of the day.

Token recommendation assistant: Help users filter and recommend Tokens that require in-depth research.

Going back to my previous point - why haven't Web2 companies made the full push and move to action-oriented intelligent agents yet?

First, the co-pilot and research assistant are already very practical. I use Grok, ChatGPT and Perplexity a lot. These tools significantly speed up my workflow and reduce the time I spend on tasks.

Second, building agents is really hard. Many startups have tried but ultimately failed to realize their dream. There is indeed a graveyard of failed projects in this area.

Objectively speaking, action-based proxies are an amazing vision in Web2. I remember the first time I saw a company like @Adeptailabs demo their agent tools - they could find homes for sale, analyze Excel sheets, and even record sales relationships.

As @elonmusk said: "Fate favors irony." Since the action model is The following two aspects are extremely difficult:

Production: converting the modelPush it to the practical application stage;

Commercialization: transform the model into a profitable product.

In the end, Adept had to choose to sell itself (and the results were not ideal).

Large AI labs are indeed deeply exploring and researching intelligent agents with the ability to act. In Q4 2024, @anthropicai released their computer usage API, allowing AI to operate computers like humans. Take a look at the demo below, which shows off its powerful potential.

While many crypto teams may have skipped AI chatbots and copilots, here There are huge opportunities for value creation. At the same time, it’s even more impressive that the crypto space has been able to tap directly into the field of mobile intelligent agents, something that Web2 startups have struggled to do even with millions or even hundreds of millions in funding.

I think the crypto team was able to do this because of the new financial infrastructure that was built in the crypto space. Everything happens on-chain, and executing a transaction requires simply pushing a piece of code.

3. Conclusion

The most pessimistic view is that intelligent agents are just another short-lived trend like NFT. To this, I would like to say: Although NFTs are less popular today, they are still an exciting innovation in the crypto field, allowing individual Tokens to possess unique properties. Secondly, how can AI intelligent agents be compared with NFTs?

Look, with the advent of AI, the world is changing at an alarming rate. Programming is getting faster, software development is accelerating, and the transfer of knowledge is happening in a more integrated way. I think in another ten years, probably no one will be talking specifically about intelligent agents or AI because they will be deeply integrated into all relevant software and become a natural part of it.

Currently, we have just touched the tip of the iceberg of AI in Web2 and Web3. Humans no longer need to spend a lot of time thinking and going through tedious processes. Now, AI can run smart contracts in crypto, dramatically speeding up workflowsprocess, creating significant practical value and added value for users.

Who can ignore this? This is not just a trend, but the direction of the future.