Host: Thank you, Professor Chen, for your sharing. Next, let us move to the final speech session of this morning. Please welcome Mr. Miguel Palencia, co-founder and Chief Operating Officer of Qtum, who will join us remotely to deliver the keynote speech.
Miguel Palencia: Hello everyone! I'm very pleased to be here and to participate with you in this way. We have always been happy to be with you, and regardless, I am delighted to be part of this keynote speech.
Today, I will talk about the Qtum space, our past work, our current position, and future prospects. Qtum was initially founded ten years ago, and today I saw a screenshot that was one of our initial announcements released in 2016, possibly over 80 of them.
We started adopting a hybrid approach, a hybrid blockchain, at the time we thought about combining the UTXO transaction model, similar to Bitcoin’s transaction model, with Ethereum's EVM, which was our original technology. It was built as early as 2016 and launched in 2017. Today, we are transforming while maintaining this very stable blockchain platform. This platform has been in use for many years without a single crash or failure, requiring a blockchain restart or similar operations, with no hacker attacks or other issues. Now we are shifting towards artificial intelligence, but not just artificial intelligence; in the two or three generations built a few years ago, we witnessed significant advancements over the past year that have driven transformation, especially in the past 12 months, moving from typical single interactions with artificial intelligence to more complex scenarios.
Let me first introduce the background. Qtum was established in 2017, although we had started building in 2016. Since then, we have made many core updates and continue to do so almost every day.
We built this idea because Bitcoin has undergone years of practical testing; we wanted to create something stable and future-oriented.
As a blockchain, we have a hybrid blockchain that combines a smart contract layer, bringing over the smart contract layer from EVM to our platform, while also incorporating the UTXO transaction model. What ties all of this together is the account abstraction layer, which is our team's invention — AL (audio). Thus, allowing two things; we did not design them to communicate with each other, namely: EVM and UTXO transaction models, to truly function.
Qtum technology has many advantages; its efficiency is very high. In fact, we might be the first team able to run a blockchain on such small devices (like Raspberry Pi). I believe we were also using a Raspberry Pi back then, and even now it is still applicable to the latest models. Since then, we have been continuously growing. So currently we have about 2100 unique staked accounts. Qtum is a proof-of-stake blockchain, not proof-of-work (PoW), but proof-of-stake (PoS), and it has many delegated addresses. We have a representative mechanism, but the direction is different; we do not want to centralize anything; it is absolutely decentralized. We have a mechanism called "super staker" whereby if you have a strong enough connection, you can let other wallets connect to you, and everyone can work together. It's like a staking pool.
But what we are currently doing relates to artificial intelligence, which is a big deal and something we are starting (things). Since 2024, we launched Qtum.AI, but at the same time, we have also launched our own hardware, GPUs, mostly running with support from our data centers. Initially, it just started as a basic AI chatbot, but later we added image generation, and today the scale has expanded greatly. I will show you this later. We have modeling layers, fine-tuning, models, hosting, and MPI infrastructure. Now we have a technology called "Qtum Router," that can virtually connect to every major provider through a single connection.
Transaction fees are also very low. Our vision is that in the future, you are not only humans using artificial intelligence; AI agents will use all tools as well. For example, as we transition to Web3, a few years ago, we hoped a new door would open where people do not have to rely 100% on banks. We began to take control of our own wallets, which is like an earlier step; in a decentralized economy run by AI agents, they do not sleep, and if they want to invest, make investment transactions, or do similar things, they can operate and analyze based on their parameters, which they have completed beforehand.
All tasks they can perform will take place on the Qtum blockchain, hence referred to as "the USB-C of artificial intelligence," a universal interface that serves all different services on Qtum's data artificial intelligence. Of course, there are additional requests, a MCP server called the open router version of "Qtum Router", all of which enable desktop automation and fully autonomous operation.
At some point, we achieved learning capabilities; these capabilities exist within the models themselves, and they actually work collaboratively, improving performance through constant training and refinement. This is very exciting because it does not rely on a single model; it can run on GPT, Kimi K3, and many others like GLM, Qwen, and of course, Anthropic. All of this is processed through Qtum Mail, handled in a very efficient and stable manner.
Last year, we launched the Qtum Alliance, a desktop AI agent that initially ran on Mac. We needed some support, and now there is support for Windows, but it is not intelligent; it is a desktop application that can serve as your assistant. Initially, it primarily focused on a few clicks and operations, along with typical chatting, but now it has begun to support voice functions. I believe the next wave of interacting with AI will not be 100% chatting; some things have to be toned down because there are too many interfaces that have not been fully explored.
For instance, 100% voice control; just imagine allowing your AI assistant to perform operations or any task, such as booking flights, making hotel reservations, simply by voice command it will execute all the information you need, even purchasing tickets for you — all of this can be done 100% by Qtum Ally.
Several other services provided by Qtum AI, with the very cool feature being 100% private, all operations related to Qtum remain on your Apple laptop, never leaving your device. All data is stored locally; any information transmitted through the API is the same, but the application itself does not send any information anywhere; all data is stored locally. One popular service is called Seedance, which might be the best video generator, with AI being the market; all of this also exists in the Qtum space. We even needed to demonstrate it. Our team members wanted to create arcade fighting games, style videos, etc., and I actually handled everything.
(Play video)
This is actually really cool; it reveals how the Qtum chain was born, the combination of everything, needing something to bring the technologies together.
The transparency on our platform is just simple text prompts, and you can get any video within minutes. This is very urgent, but you still get the full quality from the Qtum AI website, with character consistency, and indeed it will remember individual characters. If making longer videos sometimes requires multiple steps, this way the character will not get lost, nor will the facial features or body characteristics change. You ask it to remember, for example, Lora, and it will continue from there for the next duration.
"Where is the blockchain?" Actually refers to artificial intelligence, and something will emerge between artificial intelligence and blockchain itself, called "verifiable computing," where artificial intelligence itself will help validate the transactions themselves, no longer just requiring human manual operations. The decentralized infrastructure, as we used to have, does not have single points of failure, which is exactly what artificial intelligence needs. Current artificial intelligence is too dependent on data centers, these data centers run on local devices in your home or data centers we rent, and sometimes crash due to overload or limited video scenes. If the blockchain can enhance artificial intelligence in some cases, and in significant ways, that would be the centralized method of the blockchain; our goal will not affect any skills or ongoing matters, as these are determined by the artificial intelligence itself. Achieving this is not an easy task, and we are not very close to realizing it yet, but it is absolutely what we need, and we need to achieve it as soon as possible.
The "economic track" is a native payment method for computers. For example, when we rent any AI platform, subscription fees are required, or pay as you go. But typically, those session options can become very expensive, especially for some users, particularly high-demand users granted permissions similar to Facebook, who can easily spend over $1,000 a month. However, with the open Qtum Router itself, people can make small transactions to achieve similar performance, at only a fraction of the original cost. This is a service we provide, primarily including: the fields where artificial intelligence will benefit will empower blockchain, namely our traditional smart contracts and applications on the blockchain itself, which will become smarter and possess self-learning capabilities.
Furthermore, the way humans use it can largely be automated, with basic transactions, improvements, security research, audits, all of which must be completed by me, helping to make blockchain applications more secure, reliable, and stable. We might even find solutions to long-standing problems, especially at scale, as AI assistants are never off, greatly facilitating development, improving code quality, and being very secure, covering all aspects. Thus, we have witnessed the evolution of artificial intelligence itself over the past few years; initially, there were only a few tiers of chatbots, probably around 2022, DALL-E was the first to start gaining popularity; I found it very interesting, and I believe most people are already aware. But of course, ChatGPT was the first to become famous. However, there weren't many other things, other than these basic interactions where you couldn't search the web or do a lot of things, and these are things we are now doing every day. Everything is driven by us.
Two years ago, especially last year, we saw the first batch of Agents. They are more like powerful assistants capable of web searching. They can perform more tasks than the first generation. By the end of last year, especially this year, with Lobster, we began to see some changes, truly witnessing things that could execute more complex tasks or even loop through certain operations. For instance, I need you to look at this for X duration; Lobster might be the first to achieve this, although Lobster has all the performance issues and more security and stability concerns. But it showed us what could be achieved.
As early as 2023, I tested a product called "All GPT" (audio). This could be the ancestor of all such products. All GPT can create PDFs, search online for text files. It exists between all these slides, but certainly, that was a precursor to open low prices. I believe open low price inspiration came from this. But now, the situation has become more complex. For the future, it has become stronger; it will replace open cloud.
Lobster is not just an AI agent; it executes long-term tasks within loops. Because of this loop, it does not truly understand what it is doing; it is just executing tasks. The next generation will understand everything and learn, maintaining long-term memory over time, remembering everything it has done. Therefore, when the context gets compressed, it will not forget what has happened and will learn from interactions with humans, actually improving everything over time, and even proposing new updates and different suggestions that could really enhance every workflow, from programming to research, to marketing design; it's astonishing.
We are testing some outcomes that will definitely be released soon, and the progress is completely in line with expected directions. This is the future; this is how it is. It is called "perception." I believe it will be realized in about two years, especially with the improvement of large models. Every month we see the release of new models with increasingly stronger performance and context processing capabilities far beyond before.
If we compare GPT-6, Kimi K3, or yesterday's release of Oppo5.5 (audio), any product launched last year has taken on another game, and its performance absolutely surpasses anything we’ve seen before. They are approaching a tipping point where they will understand how to improve their own platforms. If you build a platform for artificial intelligence, that platform can enhance its own performance; the possibilities are limitless. Will this statue become the goal designated by humans? It indicates a better goal, initially given by humans. The learning capability will start to give us a sense of AGI. They seem to be capable of thinking, and I find that very exciting. It may be a little scary, but it's very, very exciting to see it happening. So it operates similarly to the way the human brain works. What we refer to as consciousness.
Since 2023, I started hearing about things like AGI. But that is very far off, very, very far away. However, I don’t think it will be solved by a single model. It will be a whole suite of tools, a complete platform that can be called an "operating system," on which these major AI large models can run and possess a kind of consciousness capability, which is a different type of reasoning, able to foresee circumstances before they arise. And this will change everything; everything is changing, everything we know is changing.
From cybersecurity to government applications, of course, even including defense systems. Many countries are known to be heavily investing in defense systems, namely "AI defense systems." Therefore, this is a significant paradigm shift, different from everything we have seen before, open cloud, autonomous agents, OpenAI earlier this year. Lobster has created a stir on social media, reaching about 20 million monthly active users. It has grown so large that Anthropic decided to ban Lobster, and Lobster can no longer connect to Anthropic unless using some strange tricks. They do not allow connection to describe models; you can run it through the API.
But this is because many people began using cloud services, and at the same time, this actually caused server crashes. People would ask these agents to handle a variety of tasks, like focusing on creating websites or engaging in predictive market activities, which is actually quite crazy, the speed at which it is developing.
The only issue, I believe, is a limiting factor for this type of technology, the limiting factor lies in usability. Setting up the relevant features is not easy, like Lobster or Harness Agent. There exists a huge technical barrier. Once this technical barrier is overcome and it runs in a very easy manner, this is precisely what we do with QtumAI. QtumAI will bring in more users because 20 million users a month are all technical users; future growth will be aimed at everyone, anyone, not necessarily a technician but who has some understanding of ChatGPT or has some knowledge of artificial intelligence. When they are able to access such autonomous agents, there will be significant growth. Any platform that makes this technology easy to use will undoubtedly dominate the entire market.
We are beginning to notice a cool thing, which is the communication between agents. This sounds a bit like a plot from a movie, robots talking to each other. Now we are testing internally, testing every day. They do not just converse with each other; they are truly working as a team. In the development of finance, everything you can think of, including marketing, etc. This is ultimately something these platforms can rent out, renting to anyone who needs to build something quickly, with the platform itself recommending the number of people needed based on task difficulty (for research).
Ten years ago, we were focused on decentralizing everything. Qtum itself has its own GPU; of course, we combined our blockchain consensus, which has always been open source and will always remain open source. A powerful technology like artificial intelligence should not be closed source. It should be something everyone can understand, thereby understanding the actual situations behind it.
For many reasons, such as security, privacy, and so forth, the entire token economic system, especially Web3, is certainly applicable to artificial intelligence and Qtum AI. In fact, I believe this can absolutely be called "Web4." This is the next step.
Initially, we experienced an aging economy, and this year we have begun to see it integrate into the Qtum space. Agents have begun trading on the chain, at least on the testnet.
The coolest part is multi-model coordination. So you no longer need to rely solely on a single type or single agent, but rather have a team, scaled to personal demands. It could be any number between 3 to 100. With 100 doing different things, some of them coordinating, they are responsible for reviewing every task, assigning tasks, and establishing actual workflows, like any IT company, or any business or development team; this is absolutely worth a look. Certainly, in the second year, we will see different faces. All these agents are self-learning, executing tasks, and continually improving over time.
So far, based on early testing, I have seen that this has already happened. Once it enters the market next year, it will absolutely make everything we have witnessed this year seem like "a product of ten years ago." This is certain. It will continue to evolve and develop over time, becoming stronger; many countries have already opened up cybersecurity, AI cameras, facial recognition cameras, but they are actually opening up parts of government operations to artificial intelligence. Of course, we cannot achieve this with last year's technology but must use next year's technology, that is, the technology in 2028, to see this in many major countries. For example, Dubai and Abu Dhabi have launched a large project to integrate artificial intelligence into the government system. It was even announced that steadily and inevitably, artificial intelligence will participate in government operations in the next two to five years.
Of course, some are also predicting this kind of issue that is similar to what you see here.
But let me take a second to show you the "Qtum website." Thank you!
Host: Thank you to all the guests for your wonderful sharing, and thank you to everyone present for your focused listening. The morning agenda comes to a close here. From digital currency payments and settlements to consensus and divergence in future financial infrastructure, we have seen a clear thread of the industry moving from technological exploration to institutional building. But this is just the beginning. We'll return here promptly at 1:30 PM this afternoon for even more substantial content sharing. See you at 1:30 PM!
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
