Ying Yan - The Future of RWA: DTVM's Asset-Driven Open Stack - TOKEN2049 Dubai 2025

I'm very excited to introduce our innovative opensource initiatives uh that is the smart contract engine together with AI power to set that can revolutionize the tokenization and today we are studying and the threshold of a new year in tokenization a year where traditional tokenization practices actually are evolving into the intelligent system. And this uh revolution actually is marked by the ship from tokenization into smart tokenization. And smart tokenization offers not only the flexibility of adaptability but also can support the performance boost and by the enhancement of uh uh artificial intelligence and it should also empower the asset development and management and what's more it should also support confidentiality right uh verifiability and we need to make sure that every process should be pro uh should be uh smarter secure private and uh efficient right so therefore we can see from uh tokenization into smart tokenization is now just the improvement it is a revolution in the web3 and AI era but how can make this future a reality let's first look at the life cycle uh see uh in the digital asset we see this full life cycle and uh at hook position it is smart contract right smart practical asset and applications. It is triggered by the transaction and is record every state changes and make it temper proof and transparent. Right?

What powers those smart contract? Yeah. That is the engine behind that is a virtual machine. Yeah. So we see virtual machine is very important.

It is the key to unlock the potential. But however look at today we are facing to the future. There are a lot of attributes. For example, we need adaptability, flatility, we need the performance, high performance, we need to handle the AI capability. So, we have a confidentiality issue.

The existing uh virtual machine cannot surface. So, that's why we see there a lot of project are talking about how to upgrade their virtual machine, right? And the answer for us is simple. Uh we need to uh lead the future. We need to have a next generation virtual machine and we are ready.

And today um very I'm very excited to introduce this is open sources uh opensource the virtual machine we call it a deterministic virtual machine and together with AI powered two set um so for uh I love this uh AI tool uh smart cogent because it encodes smart contract code agent it can help developer uh to solve the problem of uh smart contract developing optimization um and smart contract auditing and deployment all auto in automation there no more delays no more errors um they're just auto in a high quality right but besides the name I love I also love the cool features provided by the DTBM stack so let's see one by one the first one is performance right performance it has a deterministic execution uh based on the compilation technology and achieve very good performance result so it is EVM compatible So we compare the performance with ERC uh those tokens actually are running on EVM. So we can see for ERC20 721 uh we can achieve two times speed up but we know those uh smart contract are relatively simple. Um let's see the complex contract. So for the complex contract we can achieve up to um let's say uh 50 times speed up on execution and 23 times uh speed up on the compilation time right and the second feature is it is a secure hardware compatible for example TE right uh the technology behind is that we use the hardware acceleration this is uh bug checking right this is to improve the security and on the other hand We reduce the trust code base. We reduce size to 30% compared to warmer.

So smaller trust code base means um less dependency. That means that we can work together uh very well with those uh trust hardware and also smaller trust codebase means less possibility to be attacked. So it's very secure. And the third feature also is very cool feature to all the developers that is bes besides solidity uh we support the other file languages with pl uh which include C++ uh java go um we say rust and typescript that means if you are developer you know one of the language that you can develop web three applications this is really I think inspire a lot of innovation to all the developers right and the Last but not least is our AI power tool. Uh just to help you you just use uh natural language.

It can help you to uh generate the smart contract code and optimize the code and also audit the code and deploy to the onchain and connect to your wallet. Right? So this is really cool and you need to uh focus on what to do instead instead of how to do it. Okay. And then let let me introduce the story behind those technology.

We first look at the whole DTBM stack. It has three layers from bottom up. So we have this engine layer which responsible for the deterministic execution. So this feature actually want to make sure the smart contract can run on every single environment, right? Because this is a decentralized network and we have a standard currently proposing and they have five uh deterministic behavior that that need to be guaranteed and also through our function level lazy compilation technology that we can have very uh speed speed up uh contract execution.

Right? This is the bottom layer and then in the middle layer is SDK layer and here that is why we can support six languages. we need to solve uh every single technology problems with different languages that can make a developer easy to use right and on the bottom we have AI AI tools right and besides performance we also know this confidentiality um is and verifiability is also very important and we also extend this uh T and ZK framework for example and here we shows the ZKP framework so we just develop the ZK circuits algorithm and also develop the proving system that can work with different algorithm and also we have hardware acceleration that make the uh proving speed uh as fast as we can. Currently we are working with many cool projects for example SP1 stockware uh and ZQVM with the cool pro products that we can help them to do the ZK acceleration to 100 times compared to CPU. Yeah.

Okay. And this part is about uh uh the virtual machine and then let's go to see the AI tool. And before introduce uh what is it let's see a demo to see how if you are developer and what this tool can offer you. So okay video please. DTVM smart cogent provides you multiple agents to generate audit compile and deploy contracts.

Here are three demos. Smart cogent copiler is integrated into client to create contract projects and generate codes. Here is an English auction contract. All you need is to provide rigorous, clear and explicit requirements descriptions. Type the contract features and run.

Just enjoy your coffee. Your smart contract will be generated automatically. It firstly retrieves from our selected and sanitized code bases, extracts frameworks and supplements business logic. Here you can see the contract code in a foundry project. It strictly meets requirements.

For example, the bid increment mechanism enforced to strictly exceed 5%. Naturally, it intelligently invokes compilers and audit tools until delivering fully functional and secure contracts. Let's now audit a real world contract named pawn. Just ask smart cogent to audit this contract. Fine.

You can identify potential vulnerabilities with it. Firstly, an audit plan task is created and executed. Cogent performs a deep audit based on the audit plan. We wait briefly and then a surf audit report is generated and return. Finally, the LLM will automatically pass the audit report and contract file as parameters to the contract auto repair tool.

Now, the repair task has been completed. Here we can see that the repair tool has successfully addressed the issues. Similarly, just tell the smart cogent the name of the contract or project and they will be compiled and deployed. The builder agent automatically analyzes contract dependencies. The success rate of compiling mainstream dependencies has increased to over 95%.

After compilation is complete, the deployer extracts constructor type information and prompts the user for input. When signing the transaction, the tool actively triggers the MetaMask wallet plug-in, waits for your confirmation, and then signs and sends the transaction. Upon successful deployment, Smart Code proactively retrieves and displays the transaction data. Additionally, you can ask smart code to query any transaction from the blockchain. Stay tuned for more DTVM technical releases.

Yeah, it's cool, isn't it? Cannot wait to use it. Actually, this open for free trail. Later on, we'll show you the address. Okay, now think about you are building a smart contract application.

It just like building a sky a skyscraper, right? You need a team of people. You need a pe a team of expert professionals, right? So yes now you have a smart cogent then just like you have a team of agent AI agent for example here so we have three group of agent actually each group we also have sub agent right let's see uh the develop phase so we have two agents two group of agent right in the audit phase so we have four agent and in the deploy phase we have one agent right so by leverage large model professional data or own experience web 3 development and also the AI agent that will make everything uh automatically with uh more accuracy. So let's take uh the audit phase as an example right.

So here uh so the technology behind actually is the first one we have a professional data that we have been working on the smart contract auditing for years. So we have collect a big database for vulnerability events. So we can use the database for the fine tune and rack. And the second one is cross domain enhancement and AI agent um the multi-optimization right so for this phase so we have totally 15 agents just like you have a team of 15 expert for you to do smart contract auditing so it's very cool but the result is that we have eight times accuracy compared to today's general luggage model right and this is just the beginning we are still trying to uh make it more accurate make it use smoothly and to make it your favorite smart contract developed tool. Um and the DTDM stack actually uh is not used for one scenario.

It can be utilized by layer one project, layer 2 project, IoT devices as well as Oracle services. Just think about you have a scenario, you have bunch of real world asset. So for example you have the solar panels actually on the IoT size you can run DTBM on the chain size you can also run on DTBM and through the smart cogent tool they can move make everything secure and and smooth and then cost effective right every place if you want to have secure data and make data accurate verifiable run smart contract actually you can consider using DTVM and also actually this uh DTVM stack we already done uh the auditing by the professional team. So it is secure uh and and mature to use. Yeah.

Okay. And then uh this is uh I think this beginning it is open source and then our technology paper actually there are a lot of details for uh for us actually to to look into. So we have our challenges we are facing. So we have the the solutions we have already had and evaluation with uh different type of smart contract as comparation. So uh I really recommend that you can uh just uh skin the third cloud code and check our paper and also this is open source and welcome uh everyone to fit um test it and break it and let's together u build the next generation tokenization future and besides this uh DTVM and tool actually if you need uh uh the other web three services like uh node services auditing uh uh security chat as well as hardware acceleration.

You can also uh contact us because we provide a lot of uh basic technology and services that can uh help builders to build the future together. Okay. Thank you.