DAPPCON 2025: Powering the Agentic Web: Building Web3 Agents on Google Cloud - Marcos Montero

Transcript

[Applause] Hi everyone. I'm excited to be here with you all today. Um my talk is going to be about unlocking the agentic web. Um building web3a agents on Google cloud. Yeah. Uh my name is Marcos Montero and I am a customer engineer at Google Cloud. I work with startups and scale apps in in in a to build cloud native solutions.

I focus on web 3 AI and security. My talk will start with an introduction to the identic web and then it will be followed by the building blocks to build an architecture. uh talking about our open frameworks and services for AI and web 3. The last part will be uh reference solution architecture for DI using all of this tooling gigantic web. So what is this?

First let's start with an evolution of the web. We saw that the web one started as a read only h format where there was only content was one way and a paper. Web two added the right uh features to it and that uh was the advent of social networking. Uh the data and the systems in web two are centralized. Web3 adds this ownership feature and layer to it where dabs are built on P2P.

Content is owned by creators and the data is decentralized. Now the Aentic web expands on this with an execution layer on top. With it you get decentralized intelligence verifiable onchain and effortless and autonomous. So it helps with the UX and makes uh web3 features autonomous. What is an AI agent? Let's start with a definition. An AI agent is an application that tries to achieve a goal by observing the world and acting upon it using the tools that it has at its disposal.

What is key here is that AI agents use LLM, generative AI models. one or many to achieve this goal and they are aware of a context. So how have AI agents evolved with the advent of generative AI? We saw chatbot applications that had an LLM and a prompt. This was very basic in a way and soon people realized there were problems with okay uh those models have hallucinations and also what happens if I want to have

my own data uh on these pre-trained models of if I want to have external data sources that's when it came rack retrieval augmented generation which is a method for expanding LLMs with external data. So you can ask the models questions about them and with it you don't have to fine-tune the model which is typically quite costly in terms of infrastructure. So you can have a pre-trained model and use external data into it.

Now after that those identic systems were expanded with tools APIs that are accessible by the LMS. So you have now a much more complex agentic system. What we're seeing now is that this tooling has started to come into thinking models where there is reasoning loops in these applications and this was has been the advent of multi-ent systems systems that you have multiple aents collaborating models that call other models. Some models are textbased, some

other are multimodel, image based, audio based. And these systems up until now or up until recently have been within a network within a company operated by the same personas in a way or developers. What we see now is these multi- aent systems are start starting to talk to each other and we are building the frameworks for inter agent operability where you can have an agentic system call another agentic system and achieve a common goal.

So how are AI agents architected logically? So there are three things here. There is a runtime that has an orchestration layer. That's the agent brain. In here also there is memory short-term long-term memory. They remember what you ask them. There is a definition for the agents. What are the goals, instructions, the tool, the system prompt? What is what this agent is in charge of doing? Models.

Um, AI agents can have one or multiple models which can operate in different realms. Image, text, audio, also tooling, functions, APIs, databases. A lot of these tools are now MCP servers. So you can have an agent talk to another system with the MCP server from Anthropic which is a protocol I will be talking about later and ultimately remote agents that can be available via MCP or through all your identic framework.

The end user only talks to the aen with prompts and gets a response. These prompts can be textbased, it can be voice commands, it can be audio, video, it can be multimodel. So what are the use cases of AI agents for web 3. So I will be focusing on the architecture at the end on these two top ones. But like basically how you operate a wallet, how you do blockchain operations that can be automated by an agent. Um submitting transactions, analyzing data, managing the wallet

itself. That also helps a lot with user experience because you can use prompts to talk to these agents and you might not need to have a online or called uh or offline wallet. When it comes to interesting is also for DeFi. Um you can have things like automated trading, smart staking, analyze market news. These agents can operate h synchronously based on user prompts but also you can command the agent to run autonomously a reasoning loop or things like operate this trading strategy. um do a smart staking for me, choose the best staking

notes so I get the most yield. Uh execute this portfolio trading strategy for me based on the news uh on real time. So that's where it gets really powerful. Also we have other uses like in governance treasuries autonomous decisions and gaming and more. Okay. So talking about Google uh our AI services which uh are available in our tool and also are open and we provide third party solutions as well there.

So at Google we have multiple AI services in different layers and we provide flexibility at each of them. We have firstparty infrastructure and services and models but we also support thirdparty uh infrastructure models and services also we have our own um uh protocols and frameworks but we support third party ones. So um Gemini which is here is an agentic system that is available via app and also via the web but is also a family of large language models.

And when it comes to the agents, vertex is our platform for AI and mail where you can use in GCP and ADK development kit is a framework which is uh open source and supports our own first party models but also third party models. You can use it in Google cloud or you can use it anywhere. Also we have Jenkit which is another agent development framework. and third party frameworks which support line chain and others. When it comes to protocols, we are leading a new protocol standard called agent to agent protocol A2A that has more than uh 50 partners

already involved and this allows for inter agent communication. So it spans on anthropic model context protocol MCP analogy of it is like a USB interface. It allows you to connect things to an identic system while H2A allows the collaboration between them. ANS space which is the last one here is an enterprise solution that uses agents. Uh and you can register your agents to use it. Um this is how it all works in an architecture.

Main thing here agent engine that's a runtime uh is how you can run agents on Google cloud. It's totally serverless. It's built on top of cloud run that you might know if you use GCP and it provides all those things for you. Then in it you can use our first party ADK or third party agent frameworks if you want. We support a agent to agent protocol for inter agent communication tooling like MCP other tooling as well. our first party models Gemini or thirdarty models as well and we have features for it or servability valuation and you can register them in a space if you want to

make your agent available for an enterprise. Okay. So talking about web three, we do also have uh web3 services and I will be focusing just on two of them which I think are the relevant ones for the architecture. Starting with RPC blockchain RPC is our RPC service. It's multi-tenant and you can use that to read and write blockchain data on Ethereum using the standard API. We support Ethereum, mainet, hol sky and yeah you can use it in Google cloud it's it's managed for you also we have blockchain analytics which

is index blockchain data that we have for 25 chains of those 25 chains eight of them are indexed by us so that means we run the indexing pipeline and we uh promise SLAs and SLOs's service lay agreements and objectives for those eight chains. If you want to use this data, you can use through SQL interface or via popsup. Now, this is going to be the last slide which is all of those concepts together.

So, uh this is something I implemented um as a solution architecture. I I I design uh for a defi use case that uses all these building blocks. Uh in here the user can do things like send ETH, check the balance, what is the price of it, stake it, tell me the crypto transaction volumes for last week, but also more smart operations that you can the agent can run autonomously for your D5 portfolio. So all of this is hosted in GCP in this case and you have a root agent and two sub agents. The root the root agent is the cryptobase agent. The sub agents is a wallet agent and a trading and staking agent.

KMS, Google KMS is used to store the keys and then we have RPC nodes for Ethereum and also index data for Ethereum and others if we want to. Model armor provides generative AI gals. So every prompt and response that your agent system gives the user uh in or out is inspected and sanitized by model armor. Uh so then you can prevent jailbreak attempts which are always on the on the news and anything that is uh a content safety issue you can also tell it not to speak about competitors and things like

this. So the way this works for example user says something like this in text it could be also audio and for example it's an Ethereum so it will end up going to the RPC node and for that you will get you will sign the transaction here centx cent x and it will go there. This is an example of things. Um also you can also have um things like um buy assets, get the news, get chain data that could go to external MCP servers or uh agents.

So this is showing um ADK a development kit that's the icon but also we support it's also other frameworks are supported and this is why the importance of the A2A protocol we need to standardize how agentic systems can talk to each others in the internet or in web3 and yeah I think it's very powerful it can help with UX uh it can help make uh web3 widely adopted and it can bring automation and a lot of really really cool features to to web three.

And with that I think I'm on time. I have three minutes more. Um um here uh if you want uh we have the startup program that you can follow if you want credits to build this identic systems on Google cloud and happy to connect of course and with that I have time for questions. So any questions from the audience? Yeah. Nothing. Yeah. First you mentioned 25 chain. So uh do you support?

So we actually working on this uh hopefully and and I cannot tell you just now but we can follow after. And we would love to have yes uh nosis chain on BigQuery. Uh so yeah uh we can follow up on that after I I don't know what is the status but yeah hopefully yes go for it. All right what what's the real blocker for making it just infinite chains? Like if you have an RPC why can you not index that chain? Um well we have to run the indexing pipeline somewhere. So the way this normally works uh you will have an

RTC node and every layer one or layer two will need to to build that uh infrastructure so that the node is running and then we push that to bequery because once you have you have a public data set you just click and it's ready for you but uh it's serverless it's great fantastic you don't there's no boot up time or anything like this but that means there has to be a pipeline So yeah we someone has to build this and this is why we allowed inquery for anybody to build a community data set.

We index eight of them ourselves. We run the pipelines but also there is like up to 25 minus 8. So those are community uh chains. Any other questions? Yeah. Here uh in the last in the last slide I see that you showed that the user talks to an agent and the agent talks to two more agents. Yeah. Why don't you for example in the first agent use an LLM that would categorize the the for example the user input and then talk to the other agents because that would be faster right. Yeah definitely this is the way I build it.

So you have comp it's composible and you can use if you want the well it agent for something else but definitely you can build all this in one single agent but it's kind of the the paradigm of microservices and monolith architectures. So it it's normally better and we see this to to build aic systems that have sub agents that have differentiated task and the way I envisaged this it was like okay how about an agent that does all the wallet operations while other does all the smart forwardlooking crawling new servers and things like that to build defy strategies. So that's how I

envision it. But yeah, you can do it as you want. Thanks. Any other questions? I think we are running out of time already. Okay. Well, thank you so much.