Video: Supercharging Apex, Flow, and Agentforce With Heroku | Duration: 3024s | Summary: Supercharging Apex, Flow, and Agentforce With Heroku | Chapters: Welcome and Introduction (3.4399998s), Technology Decision Challenges (91.56s), Heroku Platform Overview (287.28s), Salesforce Integration Process (456.65s), Augmenting Apex Demo (965.72s), Scaling Async Processing (1310.135s), Carbon Footprint Calculation (1669.4651s), Extending Agents' Capabilities (1870.8251s), Use Cases Overview (2271.715s), Conclusion and Q&A (2561.915s)
Transcript for "Supercharging Apex, Flow, and Agentforce With Heroku": Hello, everyone. Welcome or welcome back to another Salesforce Ben webinar. As you're joining, do let us know where you are dialing in from. It's always great to see how global our audience is. Today, we'll be exploring how to build and deploy scalable AI applications for Salesforce using Heroku. Our session will be led by Andrew Fawcett, Vice President of Developer Relations for Heroku at Salesforce, and Vivek Viswanathan, director of product management at Salesforce. So for those of you who've been in the ecosystem for a while, you might know that Heroku quickly became a leading cloud platform as a service before being acquired by Salesforce in 2010. But more on all things Heroku in just a bit. We've got a jam packed session today that will cover the Heroku platform within Salesforce, including a technical walkthrough on connecting Heroku with Salesforce and some real world examples of how Heroku is used alongside Salesforce to achieve scalability. As always, if you do have any questions, do pop them in our q and a tab, which you can see on the right hand side. And also the slides will be shared in a follow-up email as well as a recording of this session if you would like to cover any of the key topics that we, discuss later on in your own time as well. But without further ado, I will hand over to Vivek and Andy to get us started. Thank you, Zaima. In today's tech centric world, delivering innovative digital experience is expected from most consumers. 74% of consumers expect a seamless digital experience. Even when I go to a grocery shop, I'm expecting them to know what I like best. It just translates to all consumers. Choosing the right technology is a crucial decision that companies make to deliver that experience. What companies want is their technology to deliver unique, engaging experience to their customers and do so fast with little overhead. In making these decisions, they must decide between, do I build or do I buy? Technology and tools to get the outcomes faster. The decision to build comes with added operations, management, maintenance of the tech. But the decision to buy can limit your ability to create the truly differentiated experience. Most companies need both. In addition, they need to research a vast increasing complex set of solutions like iPaaS, PaaS, StaaS, and more. They must also navigate new tools and considerations like AI and agent force. Once they have done this, they have to now find skilled personas to build and manage the experience, which is often in short supply. According to CFGF in 2023, '50 percent of the organizations face challenges due to shortage in DevOps skills. Once they decide on what tech to use, companies then face with a series of additional friction steps and decisions that they need to tackle. Let's look at some of those steps. They need to first align an overall vision and the digital experience that they want to deliver to their customers. What is this big idea? Once they have aligned on that, they have to now develop a model for that experience. Okay. They did that as well. Now, we've got to build and deploy that experience. Great. That's just one step. Actually, no, that's about 3,000,000 steps. You have to think about security checks, go through compliance, go through security review, make sure you have monitoring in place, make sure you have observability in place, you have alerts in place, you have logging, and so on and so forth. Then you've got to decide when to launch this experience. Finally, but not the least, you have to manage and scale this application. Say you have a Black Friday sale coming up. How do you make sure that your infrastructure scales? How many servers do you need? All of that comes with ongoing manpower and decision making that you need to have on an ongoing basis. The decision to build versus buy, the thousands of technology choices to choose from, finding and retaining the skilled personnel. All the friction that comes after choosing a tech can cause companies to fall short of the innovation that customers have come to expect of them. This is where Heroku comes in. Heroku accelerate your outcomes and seamlessly enables your company to create new applications that are unique to your business and your customer needs and are seamlessly built and deployed and scaled all within a trusted, secure, compliant platform. Horuku is a beloved Salesforce product and it seamlessly integrates with your Salesforce tools and services. We will see this in today's demo from Andy. Horoku enhances Salesforce and helps maintain the full view of your customers' data, expanding integrations and possibilities of what you can do with the data that is in Salesforce. Heroku is a Cloud native solution for companies looking to improve and scale and lower their cost. At the same time, modernize their past investments by deploying their applications to Cloud in an easy and efficient and scalable platform, Horoku does all of that for them. Sounds great. How does this work? How we do all this is a platform that takes care of the needs of your developers, your platform engineering, your IT, your SRE. All your observability, all your metrics, all of that is provided in one single platform. All of this is delivered in a delightful customer experience that allows developers to create apps using modern coding languages that they prefer. Your developers are focusing on building the differentiated application and customer experiences rather than figuring out what to do, how to set up the infrastructure and the DevOps pieces of it. Huddlco has been evolving its platform to meet the ever changing needs of app teams while maintaining a seamless user experience. As our Cloud landscape has shifted from proprietary tools to open source Kubernetes and AI based driven development, Heroku is adapting to ensure that it provides the same powerful, efficient platform for the modern application development. At Heroku, we believe in empowering developers, which means that the best choices are the ones that you do not have to make. The modern developer is overwhelmed with choices. It's not good enough to be a full stack developer. It's common for them to basically have the responsibility of containerization, base image updates, potentially even operating the cluster that the app runs in. Like Steeda, which is our proprietary platform, Fur is built on the core principle, maximize the developer productivity by minimizing the distractions. So that's all great when it comes to managing the infra, managing the DevOps, all of that. But the full session of today is how do you connect this to agent force and Salesforce and make that whole piece work. Step one is something that we just talked about, ensuring your apps are hosted in a secure, trusted environment. But then you have to go through all of these 14 steps to basically get your app to start talking to Salesforce or AgentForce. Step two would be to create a connected app in Salesforce and enable API access for the apps. Step three would be to get the OAuth token, get a handshake from Salesforce. Then basically if you need data Cloud, you need to make sure that the connected app has the data Cloud scopes. Then at step five, you have to store this token somewhere securely and then refresh this token at a refresh schedule. When you have done this, you have to go through a security review with your security review process within your company to make sure you're compliant. You are storing the tokens compliant in a compliant fashion, but also they are refreshing these tokens in a compliant manner. Now, all of these steps have to be done for each and every one of your apps. After this, you have to determine whether you want to use the Salesforce SDK or the Data Cloud SDK. But before we go there, you also need to determine which kind of token you want. You want a JWT token or you want a Webflow based token? Which which one do you want? And sometimes the Salesforce Core versus the Agent Force require two different types of tokens. And then, as I said, you need to determine whether you have you're going to use Salesforce or you're going to use this data cloud SDK or in the future, Agentforce API. Then you're going to write your custom code. If you are going to use Data Cloud, you need to write a piece of code that's going to exchange your Salesforce token for a Data Cloud token so that it can talk to Data Cloud. It's not just about giving the scope of Data Cloud, but you need to actually get a Data Cloud token to be able to talk to it. Once you have done this, you have built your app, you know which token you're going to use, you're going to use the SDKs, written your code, then you basically go ahead and test and deploy this code to Heroku or a platform of your choice. At that step, you will then go ahead and create a named credential in Salesforce. This is all in preparation to get this as an external service within Salesforce. If you have gone through the import process of an external service, though it shows up as just step 12, it's actually more than just one step. It's about four steps within itself. Once you've done that and if you imported your API spec, it's only then at step 14 you can start using it as part of your flow or your Apex code. Even then, the only way you can run this app is as a system user on an API user. You cannot get the permissions of the user at the keyboard. That means you cannot run the app as a sales user. What is Heroku doing to change this? Just like the infrastructure piece where we are taking care of your DevOps, your infrastructure, all of those pieces, we are giving you the same seamless experience to the developer to make sure that they can go from step one to step 14 in one single step so that they are focused on building those customer experiences and not focused on building all of these connections. We have two add ons now that basically will take you from step one to step 14 depending on the integration pattern that you want to use. What are the key components of Horeku integration service? The first one is AppLink. If you have not seen TDX, we actually announced this in TDX as well and we had a customer speak about it. Horoco App Links provide a seamless connectivity between your app and Salesforce, Data Cloud, and Agent Force. It's a seamless connectivity between the two. What it also provides is the ability to now visualize your apps within the Salesforce setups. You can see what your app contains, your API, your open API spec, what are the operations supported, what is the input output parameters, and all of that can now be seen within the Salesforce setup under Heroku apps. The good part about getting to see that is not just about the visibility, but now you can now seamlessly use it within Apex and flows as well. If you're using low code, you can use it within that. If you're using more code like Apex, you can use it within that as well. In addition to that, we provide managed security. Yes, we connect the app to the account. That is one set of connection or security that we manage for you. But there's another layer of security that we provide you. We provide three modes in which your app can run. The first mode being user mode. As I mentioned, the user and the keyboard, they have permissions, roles, hierarchy, permission sets. All of the visibility that you set up within Salesforce gets magically transferred to your Horoku app so that when you use our SDKs, which are pre hydrated to the context of the org and the user, you will only get the data based on the visibility for the user. If it's a sales user and they have access to only 10 out of the million accounts, your app, when it reaches back to Salesforce to basically query or perform DML operations, it will only have access to those accounts. It's not going to get access to all million accounts. That gets seamlessly translated. That's more important when you think about agent force actions and so on, than you have, say, an app that is basically passing PDF files that are attached to an account, but the or or quotes that are attached to the account. But the visibility needs to be specific to that quote that the sales user has access to. And as I mentioned, we provide you SDKs so you don't need to worry about whether you choose a Salesforce SDK or Data Cloud SDK or the JWT token or the OAuth web based flow. It's all pre hydrated for you and ready to go. The second piece of the puzzle is Horoku eventing. Horoku App Link is deeply integrated, shows up in Salesforce setup. You can see the open API spec and all of that. Horoku Eventing is slightly loosely coupled. It basically is one stop shop where you can manage all your subscription and publications. You can subscribe and publish to Salesforce for CDC as well as platform events in a single command. We will see this as part of our demo as well. It provides the same security and compliance when it comes to managing your tokens, managing your visibility, all of that. But it also provides industry standards like Cloud events and pop up. You can now have inline filters for your streams. You can actually create a stream where you subscribe and publish, and on the publish you can actually filter out which events go out. You can subscribe to those events using a webhook in Heroku or webhook outside of Heroku as well. So when do we use Heroku App Link or Eventing? So Heroku App Link is, as I mentioned, is deeply coupled. So if you're looking at using agent force actions or custom agent force actions with your choice of language, you would use Heroku App Link. Now when you're using looking to use flows and Apex, you would, again, come to Heroku App Link as well. But when you're using Heroku eventing, you'll slightly decouple, just want event based interactions, near real time interactions, you would use Heroku eventing. The combination of both is very powerful when you think about it. You can listen to a CDC event and then use an app that basically goes back, reaches out to Salesforce, queries data, and then performs operations directly into Salesforce. We have seen all three use cases being used by customers and we actually had Workday present at TDX on how they used each one of these products in combination or by itself to solve their business use cases. At this point, I'm going to pass it to Andy to go through a quick demo of the two products. Hello, everyone. This is Andy Forsey. I'm gonna be taking you through our four demos today. The first is going to be augmenting Apex Logic followed by scaling async processing, followed by extending flow with more compute and also extending agents with even more compute. So let's get started with demo one, augmenting Apex Logic. So we have, some Apex Logic here on the right, and we have some Java logic here on the left. Just gonna take advantage of the processing power and rich libraries available in Java and in Heroku. So what we're gonna do is use the Visual Studio Code add on for Heroku, which is brand new and GA'd last week at TDX. So we're gonna start that deployment going, and it's super easy to use. Just a few clicks. Deploying to Heroku is really easy from the CLI as well, but we're delighted to bring this extension as well as a few other features to market. So, while that's deploying, we'll talk through, what's going on with this use case and the particular code that we're going to generate. So, what we're talking about here is augmenting Apex code. So we're not throwing away our Apex code, but where we need more compute power or perhaps the power of other libraries or other languages, we wanna bring those in easily using the AppLink product. So on the right, we're going to run some Apex code that generates a a campaign. And in this case, we're gonna not only generate the campaign record, but campaign content. And typically with campaigns, you have social media cards that make your social media, messages pop. So, that's what we're going to generate with this Apex code. And you can see at the top here, we're calling, an external service, which is gonna be the effectively the Heroku code on the left here. So this code just loads a logo image and calculates some text image, dimensions because it's gonna take, some text that wants to go on the social card, the things that you want to say, as well as some image that represents, the particular logo or the campaign that you're generating. So it's all Java code. It uses Java libraries to generate the image in memory and then generates a base 64, encoded, representation of that to pass back to Apex. So while we've been chatting, this indeed has deployed. So we can see with our Visual Studio add on, it's up and running in Heroku all, without any fuss of managing infrastructure and scaling and all that stuff. Just deployed it. It's up and running. So now we wanna connect it to our Salesforce orgs. Let's go ahead and run, that first command to do that. And with the the AppLink product, you get to Salesforce connect command, which is gonna pop up and ask us for the org that we want to connect this Heroku application to. So that's gonna go ahead and, log us in. It looks like that's doing that off screen, which is fine. And I'll just completely a login prompt here, and we should see the CLI updates with that, information. Yep. It's finished connecting down here. And next, the next command we run is the most exciting command. It's the import command. So we're going to run, that. And that takes this application that's been deployed and makes it visible to one or many orgs. And, then magic can happen in Apex and Flow and Agent Force, which is the whole theme of this, webinar that we're in. So now that's imported, let's go ahead and take a look at, what's happened. So we've had Heroku under the setup menu for some time, and you can browse all of the resources and use cases that you can things you can achieve with Heroku. But now we have the Heroku apps menu. So that's gone ahead and imported, the application, the Heroku app, which actually has one or two other services that we're going to be using in this, set of demos. But we can see our render card, method that we were just looking at in Java that takes the that renders the custom social card for marketing campaigns. So super. We can see that now. And one of the thing that, I wanna draw attention to is that on the Apex page, we have generated some helper classes. So you don't have to pick your way through JSON, figuring out header authentications and all that sort of stuff. No. No. No. No. No. That's done for you. So just use these helper classes as we are doing in this, Apex code on the right here, and all will be well. So let's, keep our code run. We're gonna run that through the execute anonymous option in Visual Studio, and, that's gonna output some map in the console to the, category or other campaign that it generated. So let's go and follow that link. And, it looks like that was, generated off screen. So I'll just do my best to bring that over here, and we can take a look at that generated campaign. K. So here is the generated campaign. We see indeed we've got one attachment, which is the social card. And when I open that, there's our generated image with the text that came from Apex. So that Apex code successfully did its job. It created this campaign. It aug it was augmented with some processing and a Java code in Heroku, and then came back and created the campaign. So in summary, we had some Apex code called Java code. It used the AppLink add on to do so, and it executed its code Julie and generated that image and then responded with that base 64 image back to the Apex code, which when went on then went on to generate the campaign record and the social card. So great example of how you can take existing Apex code, call out, very, very easily now with the AppLink into Heroku hosted code, which of course can be no Java, Python, Ruby, dot net, or many other languages. So let's move straight on to demo number two. Demo number two is all about scaling async processing, which, of course, you can do with Apex to a degree, but there are certain, intentional limits that are in place when you do that. The power of doing scaling with Heroku and AppLink is that the compute is yours. You it's provisioned for your purposes and will execute, when you need it to. So, let's go ahead and take a look at that demo. And for that, we're gonna use opportunities, and we're going to pretend that we need to generate a lot of quotes on mass for that. And, typically, pricing calculations, do run into a lot of complexities for large enterprise businesses as it happens in terms of regional and defaulting and calculation rules that, are all needing to be applied. And, we've gone ahead and deployed the Heroku app in this case. You can see it's using the Heroku events add on and also the Heroku app link add on, previously knows Heroku integration. We've also got our Redis store because we're gonna be doing some, some queuing to pass the work onto Heroku workers for extra scale. So this is the Heroku application code that we we used and we deployed and, executes a batch of code. It runs SOQL queries to get hold of, the opportunities, and then it builds the quotes in memory and calculates their pricing and then inserts those quotes. And right at the end, it'll send a platform event back to say that it's done. Now all of this code is running in what we call a Heroku worker, which can be horizontally and vertically scaled, as you desire it to. So now that we've seen that code, let's actually give this a go. So I'm on here on the list view page, and on the left hand side, you can see, the logs from the Heroku application just so we get a little bit of real time view of what's going on. And then keep an eye on the bell icon because, once that processing is finished, then we'll see all of those, we'll see an an alert appear. And that alert would actually appear also on your, Salesforce mobile device as well. So now that we've gone ahead and changed those, items, that's click save, and Salesforce will go ahead and save those. And that's gonna generate a platform, change data capture event, which our application is listening to, and it's going to then start processing all of those, events. So indeed, we can see that. It it paused for a while, you'll notice, because it wanted to make sure it got all of the events for the one transaction. So it does some batching there. And we can see it generate 50 quotes. And sure enough, there is our notification confirming that the processing has completed. Wonderful. Okay. So, let's, look at one of the quotes that's generated. You can see all of the quote lines representing the original opportunity line items that were visible. And, yeah, that's representative of the 50 other the 49 other quotes that were generated. So what just happened in this particular demo? Well, we invoked, the Heroku logical Java logic in this case through a change data capture event. We didn't invoke it explicitly as we did in the previous demo. And the eventing add on received that event and passed it on to our Java code. And in this case, we used Redis as a queuing mechanism, which is recommended when you're horizontally and vertically scaling that you have some mechanism. Not only does it help disperse the work, but it helps avoid duplicates and all sorts of other things that can manifest if you attempt to do queuing on your own. So, fortunately, Heroku has a key value store, that you can do that with. So the processing then was undertook, and we used the Salesforce APIs to write that quote the quote generated quote data back, to the org. And then finally, we saw that event that notification appear. And that was because we sent a platform event back, and the org had a flow that was attached to that platform event, and that flow used flow logic to, display that nice little, notification that we saw in the window. So, there are a few other things you can do with Heroku eventing, very quickly just to say that multi orgs are thing with both with AppLink and eventing. So particularly with eventing, if you have multiple events, you can direct all of those events to one application in Heroku and then maybe, do some automation around those events and the data and then update a singular other org. Maybe that's your headquarters org. We also support data cloud as well. So you can use eventing to channel data and information insights from perhaps web apps you've built with Heroku into data cloud. And, we also use webhooks as the way in which we subscribe and handle the events. In fact, that demo was a Heroku webhook that we implemented. But there are webhooks that preexist, in Google Apps and if this then that and GitHub itself. So lots of possibility for orchestration of events and event driven architectures here. The one pattern I wanted to highlight as well, which isn't event driven, is very similar way of doing asynchronous scaling, is that you can explicitly invoke some Java code, which looks very much like what we saw, in with batch Apex, for example. Give it a query. It'll process that and then, return all that information. The difference here is that you're not event driven. You're maybe maybe working for a week or a month. Maybe it's in a business job that you need to run. Personally, I prefer the event driven one in most cases where business requirements work with that scenario because you're constantly processing data all the time rather than sort of batching it up. But just to say that generating in a sort of more batch orientated and processing a way, does does work just as well here and has has the architecture diagram to to show that. So let's move on to demo three, extending flow with more compute. So in, this scenario, we've got Astro Airlines, and Astro Airlines manage all of their flights in, Salesforce and all their bookings as well, and they have a detailed record that represents all of the flight details. What they wanna do, though, is enhance this view for their staff and in case where certain customers ring up for requesting information with the carbon footprint of all of their flights, which can be a very complex calculation, taking aboard all the cargo and all the, passengers on board and all sorts of other, parameters and variables. So Heroku is gonna be used to create a visual, in fact, a flow that we're gonna put on this page to provide that information. So here is the logic, and, this again is using AppLink. We're not gonna do the import and connect. We've done that already. And, it's running here a SOQL query to get hold of the, airline, the flight information, and then subsequently run another SOQL query to get hold of the booking. And in this case, it's not updating anything in Salesforce. It's just calculating that information and then returning it back. So let's take a look at what that looks like in, in the UI here because, we're gonna have have, our page edited. And this page has already got a flow that we've, pre created, that we can see this immediately working. I'm gonna drag the the flow component here, and then here we have our, display carbon emissions, and that takes the the record IDs that you can pass that on to the Heroku app. We're gonna save that. And then when we go back to the page, we'll hopefully see some additional information about our flights. And, there it is. So now we have the total emissions, the emissions per passenger, emissions per kilometer, and some information about how it was calculated. So let's take a look at the flow that went into doing that because this is another way in which you could invoke Heroku code from flow as well as Apex as we just saw previously. And, this is the calculate carbon footprint, action. And I didn't do anything to create this action. It was here because we did the app link, and the imports, and that technology automatically creates all of these different actions and all the processing that we've done. You can see here the screen flow, which is, you know, displaying what we we just saw in the detail record. So that's how the flow was built, very, very simple standard flow mechanisms, and, easiest picking the, the action, feature here and then just looking for, all of the different, actions that have been imported through Heroku in the page, rather in the flow builder here. Okay. So, what just happened with flow is we, had a flow, and we used some flow logic, to call the Java code app link. Julie passed that onto our Java code. And then in this case, we use some SOQL directly in that Java code to call back to the org to get some more information about the flight and the passenger records. And then, of course, the calculated data was returned, and we saw our information by virtue of the flow, the screen flow that we created to do so. Fantastic. Okay. On to our final demo, demo number four. Here, we're extending agents with more compute. And in this case, we have a self serve finance agreement, finance agreements, and PDF, solutions. So in this case, we want to generate, some financial, agreement information, also a PDF that captures that information for later reference. So here we are once again in Salesforce and in Kowa cars this time. Kowa cars have decided that they're gonna manage all of their vehicles in Salesforce, all of their contacts for those people that want to approach them for purchasing cars, test drives, and indeed any finance agreements to determine whether they can afford vehicles. So they've decided to use Agent Force to drive all of that. We're gonna show the the experience that, drives that. But, imagine there's a as a, Koa Car's website, and they have a chat agent on there powered by agent force. So we're gonna build some actions using Heroku that drive that finance agreement process. So let's go ahead and, see the Java code for that. So once again, we're obtaining a connection, to, the Salesforce org allowing us to run SOQL queries and perform various, calculations. And then we're also using a library from Java, a popular library. It's the, ITX iText PDF library that allows us to easily generate PDFs. And once we have the PDF, we're also gonna attach that, to the contact record, using the content version object. So in this case, we're returning some information just to the agent, but it's also, for future reference, gonna create a PDF that gets attached to the contact. So let's go ahead and take a look and see what that looks like. So the first thing, though, I wanna showcase is the have while we have imported the Heroku action, I would still need to create an agent action. But don't worry, that is super, super simple. So we can actually go to, the drop down here, and it's already one for Heroku in the create agent action. And we can see our calculate finance agreement is there. So let's go and click next on that. And you can see also if you notice when I'm browsing the code, a lot of the comments have come through from the code or annotations, which is really useful because agent force, really values that textual human readable description to know whether it can invoke your action, what it will do, what inputs it requires, and what outputs it requires. So do keep those comments coming in your code, folks, not just for good, healthy code bases, but also AI agents really value them as well. So let's, go and filter. Let's go and make sure we show this action, response without interested in any of the response types of this demo. And now we've added our lovely, action here denoted with an orange icon this time. And we'll go into our, agent builder. In this case, we're gonna demo it from agent builder. It's good for demos to developers and architect audiences because we all get to see what's going on behind the scenes. So there is our action that's gonna add it. And seems to be a little bug out with agent builder right now. The actions do all still exist. For some reason, it just clears the list entirely, but they do exist. So we're gonna go on now and, try and ask for that finance agreement. So we're gonna say, I want to buy a car with a thousand dollar down payment, a maximum of 5% interest rate over three years. Can you give me a competitive financing estimate? So it is going to do that for me, but we do need to make sure that, more more details are provided because the request, the Java code, in in this case, wanted the contact and the type of car that we're interested in. So, we're going to give it, some more information. And, to start with, it wants to know, what type of car I'm interested in. So let's, let's give it, that information. I'm interested in the ZIG model three. It's a sporty version. And, next up, I predict, is gonna want to know our who we are, our email address. So let's, let's do that. Okay. It looks like, yeah, I do want to proceed, for a finance estimate. So I'll say yes. And, let's see if it finally asks for our email address. Yes. It wants our email address. Sometimes AI does ask for things in slightly different orders. So, we're going to give it the, johnsmith@kodi.com, who I'm pretending to be right now. And, I believe at this stage, it should have all it needs. It knows the car. It knows who I am. And, eventually, we should see it call that, Java code action. So there it is. We can see, the down payment. We can see the information that's been calculated. Fantastic. So what just went on there? Well, we asked, agent force if we wanted, that we wanted to buy a car and the down payment and information was given. It asked us about the vehicle and who we are, and then it went on via AppLink to call agent, call Heroku. And the Heroku code was implemented as we just looked. It does SOQL queries. It generated the PDF and attached that to the quote. And it also, sent some query to information from the Salesforce org as well as calling some external services. So just like Apex, Java code or any code in Heroku can definitely leverage external services as well. So finally, we got our calculated data back from the action, and, of course, agent force, Julie showed us that in, the the conversation window. Fantastic. Okay. One thing I wanted to remind before I close the demos out is that, we will provide a a QR code to our AppLink pilot at the end of this session. And within that will be a tutorial to allow you to create your own agent actions with Heroku. And, this tutorial is is shown on screenshot here. The key thing is it shows images as well. I kinda think that's very cool, so I just wanted to highlight that particularly that it doesn't have to be textual response. You have the power of many programming languages and libraries here. So go to town on the type of images or media that you want to create because, it can be embedded within, the agent force, conversation window as well, not just text output. So thank you very much. On with the rest of the presentation. Thank you, Andy. Thank you, Adi. Hello, everyone. Okay. Thank you, Adi. So when we're talking about Horoku App Link and Eventing, these are some of the use cases that we've seen customers implement. Starting with trigger optimization. We actually had, presented this as TDX as well by because they had an org that's about ten years old. And as orgs grow older, you'll notice that many of these triggers become heavier and the objects become heavier as well. That save takes a lot more time than it used to when you first came up with the org. That essentially impacts your user experience. The case takes longer to save. The accounts take longer to save. Opportunities start to take time. That essentially impacts your overall productivity from your user base. So what Bugda did was they offloaded some of these trigger operations, especially the ones that can be asynchronous in nature, to Horoku. Now in Horoku, you actually, it runs synchronously. So, yeah, even when you offload it using CDC or platform events, though it is asynchronous, in the Horoku environment, it runs synchronously. It doesn't go through MQ or any of that. Once it comes in there, you can use AppLink to pull data, additional data, if you need. As I said earlier, you can run it as a user mode so it maintains that user context. Then you can write it using a DML operation right into Salesforce or you can push a platform event as we saw in Andy's demo. If you expose Apex REST service, you can actually call Apex REST service as well. So essentially, it helps you optimize your org for performance so that your end user experience improves. We also saw, you know, Workday changed their batch batch job implementation. Essentially, they are a job that runs every four hours to basically process information that that are coming in, activations for their customers. They were able to do bring this down to near real time because they were listening to a platform event or a CDC event, using AppLink to process that information and write back to a custom object. Essentially bypassing the need to batch all of these. As I said, on the Heroku end, we don't go through an MQ, you can multi thread it, so you can actually gain both vertical as well as horizontal scale if you need to. Custom agent force actions, you know, Andy did a great demo of how we can extend it. We have customers who are trying out different use cases. But if you are in the healthcare space and you want to bring in your own custom libraries to say, look at x rays or bring in Horoku inference models to basically extend the data, that you are having in Salesforce, you can do that as well, with custom agent force actions and Horoku. Flow and workflow optimization is very similar to triggers. As your org grows, you have a number of triggers and flows that you want to offload, and these become as microservices. Right? You you don't need to surface just one org. You can service multiple orgs. Orgs. With that, this is some of the common use cases that we have seen, PDF generation. PDF generation need not be just for one org. You could create a microservice and it basically can serve as multiple orgs because you can use open source libraries like Puppeteer and just use it as part of your code. Within 40 lines of code, you now can generate PDFs for your org, which has been asked from many of our customers how do they do PDF generation in Salesforce because we want to be able to do this. This can be multi page PDFs. But it can also be PDF processing. That is, you can basically pass through a PDF and then pass it back to Salesforce. It may not just be PDF, it could be image processing to pass through an image using your custom inference models or bring in your own models, and then pass that data back into AgentForce or Salesforce to basically be able to provide your end user more information. These are examples. We also have image processing. We had a customer who wants to resize these images mainly because they have it on their experience page. The page takes a lot of time when you basically have large image files. They send it to Heroku, resize it, optimize the resolutions, and then we basically take that and push it back. Now, if you think, and Heroku is HIPAA compliant, so you can actually use it with your healthcare Cloud as well and bring in data from health care. It could be images of health care and so on to process those images as well and send that data back into agent force or even your core sales force. So if you're interested in AppLink or the eventing pilot, this is the QR code. Please sign up, and we are more than happy to get you onboarded. And we are looking forward to your feedback on how we are going to use these and other designs and architectures that you would like to use it with. And and now open for questions as well. Amazing. Thank you so much, Vivek and Andy. If you do have any questions, do feel free to pop them in our q and a tab, and we'll cover them. We did have a couple of questions come through. The first one is, when is AppLinked available for GA? We are targeting early July for GA. But safe harbor around this, but early July or, towards the February release of the upcoming release of Salesforce, we should have it cheap. Amazing. Thank you. How can I get started with Heroku and AppLink? So as I mentioned, scan the QR code, and we will be more than happy to have you as part of the pilot program and get feedback from all of you. And in terms of Heroku, it's also available as a sign up. If you go to the Heroku website, you can get yourself Heroku account. So it's a little as, like, $5 a a month for some compute and data, or you may already have Heroku in your business. So check with your your, leadership to see if Heroku is part of your Salesforce purchase and if you can get access that way. So two ways in which you can access Heroku. But, as Vivek says, that QR code will give you a fully furnished, org with the, right permissions to use the AppLink pilot. Excellent. Thank you. What programming languages can be used with AppLink? Fantastic. Well, AppLink is part of Heroku. So Heroku is what's known as a polyglot platform, and that means any languages Heroku supports. And in fact, those which the community also have made Heroku support. We just launched .net, for example. But, while we use Java, it has a good affinity with Apex. You can use Node, you can use Python, Ruby, and many other languages. Vivek, I believe you're also going to provide SDKs in certain languages as well. Right? Yeah. We actually launched that Python SDK as part of PDX. I'm very interested. Make it easy. Excellent. Thank you. Another one here. When would I use Apex versus other languages with Heroku? Good question. Well, hopefully, the demos has given you an insight as to the power of both leveraging the languages themselves, maybe your skill sets or skill sets in your organization exist in other languages. And that means then your Salesforce projects can make use of a broader pool of skills and individuals to build Salesforce products. So there's that. The libraries that exist within those other programming languages are are rich and diverse. Esau is generating PDF in generating images. And also the compute that exists when you're using those languages in combination with Heroku, it has a dedicated compute for your applications, which means as Vivek highlighted with Workday's use case, it'll execute those workloads, that code when you want it to, and then you can scale horizontally, which means you can handle more throughput. We can scale vertically, which means you can give more memory if you need more memory to your code. There is literally an elastic scale when it comes to using Heroku and those code bases. So it really comes down to familiarity of the languages, the libraries, skill set availability, and also, really importantly for a lot of our customers that are adopting this is availability of that elastic compute. And, all of this comes with Heroku without the need to understand any infrastructure. We really wanna stress that part of it as much like every other part of Salesforce's portfolio. We do all the hard stuff for you. So as you saw in the demo, just click the button to deploy your code or the CLI if you're using that, and we'll hydrate all of the internal AWS services, networking, configuration for you. And with AppLink, all of the connectivity is done for you. So, it really is the best place to add more code to the Salesforce portfolio no matter what client you're using, and do that within the Salesforce's, Salesforce trust boundary and infrastructure. So, yeah, lots of compelling reasons to consider using, Heroku for other languages, if those skills exist and the use cases make a good fit. Certainly, Apex and Flow very much part of the the toolkit. You know, that's that's used the right tools for the right jaw, but, we're really excited to bring Heroku this close into the platform to make it, very transparent in how we can use it among those other tools. Great. Thank you. Another one here. Does import process as a step in DevOps cycle? I think the question is, I'm gonna jump to probably I'm the person who's asking, can we automate the import process during the CICD pipeline? And I'm I'm hoping that your GA feature for headless is gonna help us there. Yes. We will have a headless version of it, so you should be able to incorporate that as part of your CICD pipeline. Sweet. Let us know if I misinterpreted the question now. Whoever answered it passed it, but I'm glad you did. Yes. We'll be able to automate the CLIs. Perfect. Thank you. And is AppLink considered more secure for AppExchange ISVs? So AppLink is so as we go GA, we may not be supporting the packaging aspect as yet, but post GA, we are looking to support that. And we are working with the ISV and AppExchange team, to to make sure that it is secure, it is considered secure, and so on and so forth. So, Don't quote me on it, safe harbor around it, but we are working with that team to make sure that it is considered secure. In addition to that, we are actually working to get AppLink as part of the HIPAA compliance as well. Both of those aspects together should make it far more secure and appealing, to customers, especially for packages that are using AppLink as part of it. Amazing. Thank you. Any other questions before we wrap up for today? I'll just give a couple of moments. Thanks so much for everyone that's joined and, giving some great questions. I really enjoyed presenting and listening to what you've got to say. Same here. Thank you, folks. Perfect. Okay. I will wrap up the session now. Well, thank you, Vivek and Andy, for, your time today and for sharing these insights. It's been great to learn a little bit more about, integrating Heroku with, Salesforce. And to reiterate what Vivek and Andy have said to everyone in the audience, a huge thank you for joining today's session. If you will like if you do if you do want to revisit some of our key points or share them with someone you know, then a recording of this session will be sent in our follow-up email along with the slides so you can, yeah, revisit in your own time. Thank you once again, and, hopefully see you soon in our next webinar. Thanks, everyone. Thank you. Bye bye. Thank you. Bye bye. Bye.