Workflows to Connect Blockchains With the Real World Solange Gueiros | Chainlink Labs
Ethereum Denver·Mon, Mar 9, 2026, 12:00 AM
Speaker
For blockchain to be useful in the real world, the systems should work together, in workflows. This hands-on session will demonstrate how CRE orchestrates workflows in TypeScript that connect blockchain state to the real world.
Transcript
All right, welcome back. I'm super excited to introduce our next speaker, Solange Gayos, developer relations engineer at Chainlink Labs. She'll be talking about workflows to connect blockchains with the real world. Please welcome Solange.
Thank you.
Hello everyone. So let's start it. Uh I think this danger uh was so different for me the way that I miss seeing uh AI now even it was different from conference in the last of the the last year the end of the last year and yes AI is changing the way uh we built projects and the smart contents I prefer this yeah and um the way we connected blockchain and the external world should change as well this is my point. This is me, Solenigeros, de education manager at chaining clubis. These are my contacts.
Feel free to go to my website. I have presentations over there since 2019 in Portuguese, English and Spanish. I am Brazilian. This is the presentation. I have some care codes inside that can be useful for you later.
So, if you get the this care code, you have everything. One minute for you. Yay. Okay, one more. Good.
So, blockchain networks are still isolated environments and we have oracles to connect them with the world. The chain link oracles has uh have a lot of uh his services and a lot of connections to data sources, external resources like I have here, temperature, my watch, satality, weather, sports, all the financial information and real estate. And we do all of this using Dons, decentralized oracle networks. And this is so important because if you have only one node, only one computer, only one entity grabbing the data outside of the world and bring it to you to use inside your smart context. This can be compromised because it's only one node doing that.
So if you have this using a decentralized or a network, you must have some kind of consensus to bring this information to the blockchain. And this is so important because you know that when it's inside the blockchain is immutable. So let's take care of this. How will you bring the information to the blockchain? But this kind of connections are not so simple.
And imagine that current financial systems like AML, like a DVP, uh deliver uh versus payment and transfer aent uh connecting offchain data in general. collateral. All of this connecting with public blockchains and private blockchains. This is really not so easy. We need a different way to connect blockchains with the real world.
And I'm here to present this for you. Let's start thinking about workflows. And uh this kind of workflow, a trigger and cowbecking model. You have a trigger and this you fire something that will be your call back and normally the the callback is some actions and some outputs. This is the chain link runtime environment.
The CR C has exactly this a trigger and we have the C handler that take care of how the uh run is structured and then the callback. What is each part of this? So what's the trigger? The trigger is when an event that starts a workflow execution can be a time is scheduled like a we call it the chrome job or any onchain event from smart concret or any external API cow. After that we have the cow back okay when and now I have what I should do.
This is your business logic actions and outputs where you invoke capabilities and workflow have capabilities. What is this? So capabilities are the building blocks. I like to think like pieces of Legos that I can combine in the way that I need to my business. can be a chain reader or a chain writer or an API call to get some information or to save some information outside any kind of system integration.
When I talk about chain reader and writer, this is the interaction with the blockchain itself. And remember that I told before the C render is the glue that connect a single trigger with a single cow back and the C handles external data compliance crosschain workflows and any system integration. How can I use C? How can you create this workflow? You can use the SDKs to to write the workflows.
Uh you are developing this using go or typescript. These are the language. It means that it's not that I'm creating only the smart content on solidity but I I'm using other program language to do that. And all of this is secured by chain. Uh this is an example in go for you how it works like I should define what's my trigger and then I'm calling some function that will be the callback functions with all the steps that I'd like to execute.
Uh important to say that uh each trigger file starts a fresh and independent and stateless execution of your cowback. So they are uh like a closed parts and uh remember that in the beginning I talk about the domes the decentralized oracle networks we have this all the time and we should have consensus in the dons and what kind of consensus how is this when you think about blockchains the the traditional blockchains consensus means that we have the validators and they should agree to add the blocks to the state of the chain to add transactions to create blocks. But when we talk about the C consensus is more than this because it's not only about the blockchain. It's about verifying every part of our workflow like the external data fetching the API responses that we talked before the offchain computation the reads and writes to blockchains. This is the blockchain part and any external integration.
And we have different ways to have this consensus. we have this aggregation types. Uh we should imagine that we have five nodes and they should be identical to bring the the same response. If I do an API cow, imagine that I'm calling I'd like to know a balance of some account in a bank. I have five nodes that will call the same API to get that balance in that account.
They must have the same result. This is a identical consensus. But if I like to get the price of eating USD for example a data fit um the these different nodes can grab this price in different places like a centralized exchange a decentralized exchange you'll be totally different. So we should have a median like a average uh volume base something like that. Also you could have some kind of consensus aggregation with the beginning like a common prefix or a common suffix.
All of this you being a developer can define when you are creating the consensus in all the parts uh in your workflow all the actions and the outputs. So every result your workflow receives is verified not just the onchain operators operations and oh okay come back come back come back yes so you can connect any system any data any chain financial systems public blockchains private blockchains and this is the CI uh you can integrate new workflows and capabilities without break existing processes systems. I think this is a big challenge because blockchain always lives inside the blockchains and uh how many systems there are before blockchain and there are outside that we should connect to have more trust. Uh this is a good way to do that. It can accelerate the development, the integration and time to market for the onchain applications and you you build only once and run everywhere because you can run this in any public or private blockchain and this is compos the the blocks build deploy your application and some use cases for you uh stable coin issuations or tokenization of the RWAS is uh delivery versus payment a DVP onchain data distribution or this is so important now I miss seeing more development on that now the private and confidential smart contents and workflows uh here about the conf confidential compute this is uh a hot topic now as well and this is this is the link for the C the all in one secure orchestration layer they resources to you to learn more to start to use this the presentation again because if you didn't get the uh all the links this is the way that you have all the links and this is me.
Thank you so much
Automatic transcript — names and jargon may be misspelled.
