Salesforce, Inc. (CRM) Earnings Call Transcript & Summary

Hello, everyone, and welcome to our webinar about 5 Tips to Get Started with Agentforce and Data Cloud. Now I've got to say I'm very excited to be here for this webinar because there has just been so much interest in this topic. Everyone wants to do something with Agentforce, everyone wants to do something with Data Cloud and everyone wants to work out what those steps are to get started. So I'm joined here in the studio with an amazing customer, from Allegis Group, we have Saksham. Saksham, welcome, and thanks for joining us. Today, we're going to be talking you through these 5 steps, for getting started. And we're going to take some real life examples from how Allegis Group has achieved this. But before we do that, we're going to have an introduction to how Data Cloud and Agentforce work together, and we're going to see a demonstration of the true power of those platforms. Now for that, I'm going to hand over to Reenu Thomas. Reenu, I'd love for you to walk us through Data Cloud and Agentforce.

Now before we kick off, I'd like to quickly introduce myself. My name is Reenu Thomas. I'm a Senior Solutions Engineer here at Salesforce. Been with Salesforce for about 5 years now and in my experience of what the customers and clients across the ASEAN as well as the ANZ market across different industry verticals as well. And today, I'm very excited to showcase some of the incredible innovation that our platform has to offer. And the goal of today is also hopefully to get all of us here on this webinar excited about it as well. Now before we dive into the webinar, we'd like to begin by acknowledging the traditional owners of this land on which we meet today, the Gadigal People of the Eora Nation and recognizing their continuing connection to land, waters and culture. And we pay our respects to elders past and present. And of course, today, we'll be going through all of these incredible innovation and capabilities. So just to note, please do make your purchasing decisions on products and services that are available today. Now here's just a quick outline of what we have in store for you over the next 1-hour of this webinar. So firstly, we'll be going through an overview of Agentforce and how our agents can be supercharged with the power of trusted data from Data Cloud. Now shortly after we'll be jumping into a live demonstration of the platform and this is where you'll have the chance to see Agentforce powered by Data Cloud live in action. Following the demo, I'll be diving into a valuable fireside chat with our esteemed customer, Allegis. And in this session, they'll be sharing 5 key tips for getting started with Agentforce and Data Cloud, offering real-world insights that you can also apply in your businesses as well. Now finally, we'll be concluding with a live Q&A session. This is your opportunity to get your burning questions answered by our experts. Now before we jump in, there's a few housekeeping items. So firstly, this webinar is available on-demand shortly after we finish. So you'll receive a direct link via your e-mails within the next 3, 4 hours. Now secondly, please feel free to submit your questions at any time using the Ask The Presenter widget, and we'll try our best to answer and address all your questions during the session. And of course, if you do need a more detailed response or if we run out of time, your dedicated account team will follow up with you directly. And last but not least, your feedback is very invaluable to us. So please take a moment after the webinar to rate your experience and share with us on how we can improve. Now with that, let's get started. The one last slide before we get into it, but a very important slide I'd say, because we'd like to just take this opportunity to thank all of you for taking the time to be with us here today for our webinar. We really couldn't have done all of this without the support of our customers, our partners and our trailblazers. So we really appreciate each and every one of you. Now on that note, we'd like to make every moment count for this webinar for you. So let's jump right into it. Now most of us here in this webinar might have already heard about exciting new innovation of Agentforce. So again, only on the Salesforce platform with Agentforce today, we're able to bring together humans, agents, data and CRM in one trusted platform. But of course, like any AI system, no agent can work without the right data. And that is why, as you can see in this visual on the slide, Data Cloud sits at the heart of our Agentforce. And just for those who may not be too familiar with Data Cloud, let's double-click into that. Data Cloud is our hyperscale data engine that sits inside the Salesforce platform, which allows you to unify all of your data without having to build complex data pipelines. And this also now includes unstructured data like PDFs or audio/video files. Now with this unified and harmonized data, it then allows you to build and leverage trusted and accurate AI solutions, and it also allows you to then take automated and precise actions across the Salesforce platform. Now tying this all back into Agentforce. Data Cloud is central to Agentforce's capabilities because as I've mentioned earlier, agents are unable to provide precise and personalized actions with our first being grounded and enriched with the right customer data. So as you can see in this visual on the slide, using our robust connector ecosystem, you can ingest data from more than 200 systems and applications. Now once you've ingested this data, you're able to then transform and index the data, creating a foundation of unified data and insights for Agentforce to tap into. And then through the power of RAG and Data Cloud's no-code retrievers, you can then bring any data directly into prompt templates and flow automations, enabling Agentforce to access essential details from your structured and unstructured data. And despite being no code, users can still tailor the retrieval process by defining filters, ranking algorithms and specifying the desired level of detail for the retrieved data. And Data Cloud's no-code retrievers provide a next-generation experience unlocking accessibility, faster development time and improved data quality. And more importantly, all of this takes place through a governed and secure platform, giving you the ability to manage data, access, security and compliance natively. Now let's see the value of agents powered by Data Cloud live in action with a real customer example. So some of us might be familiar with Saks, a leading luxury department store chain based in America. So without further ado, let's jump right into the demo. Now we've all been there where we've bought a nice outfit from a store, be it a nice dress or a nice suit for a special occasion, only to find out the next day that the outfit has gone on sale. And I'm sure we all feel the same pain here. Now assuming that I'm a customer at Saks, and I want to reach out to see if there's anything I can do about this and hopefully get a price adjustment on my item. Now this is a common request, but it's a pretty complicated one to resolve. It requires multiple steps for a service rep to look through multiple documents and systems to get back to the customer and definitely too complicated for a simple chatbot to respond with a helpful answer. But of course, this is nothing an agent powered by Data Cloud can't fix. So let's see how data powers the agent experience and allows agents to take actions to answer more advanced and complex questions like these. So now let's get started with building out the agent experience. Now the first step in this process is making sure we've got the data that the agent needs to answer this question. So what we're going to do now is we're going to switch into the role of an admin at Saks, and we're going to jump over here into Data Cloud. Now here in Data Cloud, as we can see, we have our CRM org connected, which is green, and that is where we get our basic customer data like our contact information. But as you can see, we don't have our order data yet, which is a very important data source for our agents to reference and take actions based on. And currently, for Saks, that order data sits in Snowflake. And so we need to get that connected. So what we're going to do is we're going to click on new here to establish a new data stream. Now clicking on new, we can see all the various connectors that are available to us. So for example, we can see all our different native connectors within the Salesforce ecosystem. If we scroll down, we can see all the different data platforms that are available as well, like Databricks as well as Snowflake. And with Data Cloud, you can now access this data with zero copy. And on top of these connectors, we also have all these various incredible connectors that work out of the box provided by MuleSoft. So really, there is no data that cannot be connected into Saks. Now as I've mentioned, Saks is currently storing all their order management data right here in Snowflake. So let's get that connected. And what happens in the background when we connect all these different data sources is that we have a set of harmonization and identity resolution capabilities running behind the scenes to map our new data streams into all the data that we have in Data Cloud already. So just clicking here on Next, this is a visualization of that data. As you can see, this is essentially a network representation of all that connectivity of that individual, the customer and their purchases, for example, and all of the data that I've just now connected from that external source from Snowflake. All right. So now that we've got the relevant data connected into Data Cloud, let's now activate this data using Agent Builder, where we are able to build out that agent experience. So here we are in Agent Builder. And this is where we give our agents the ability to take actions and address different inquiries with our topics and actions. Now in this case, we've already set up our service agent. And as you can see, we've already prepared a topic over here regarding pricing questions. Now if we were to double-click into the topics actions. Going back to the goal of our demo, we wanted this agent to be able to help our Saks customer with being able to process pricing adjustments and refunds based on the product eligibility. So I've done a bit of a head start here and have already created this action called "Generate Price Adjust Response," which is essentially a prompt to answer the customer's question about making a price adjustment. But I'd like to edit this prompt to make it a little bit more intelligent. And so for that, we're going to now head into Prompt Builder. Now to me, Prompt Builder is one of the most exciting innovations that we've created over the last few years. At Salesforce, we've had a track record of low-code innovations. And the whole goal of it is to really make difficult technical problems simple. We've done it with workflow automation, with Flows, with integrations with MuleSoft, and we've done it again with Prompt Engineering. And before we get into the Prompt itself, I'd like to highlight a few reasons why our customers love building prompts with our Prompt Builder. Now firstly, on the right-hand side here, we've got our Models tab, which allows you to choose whichever model you would like to use under the hood of your prompts. And especially these days it seems like new models are being rolled out every single week. So here you really have got that flexibility and freedom of choice to choose whichever model you'd like to power your prompts. Now secondly, when it comes to our prompt builder, its Prompt Engineering Salesforce way, which is the low-code way, making it much more scalable for your business. So why don't we go ahead and get started with our prompt here. And to save us some time, I'll copy and paste my prompt here. And as you can see with the prompt, there is no coding involved. I've simply just written down what I wanted to do. So for example, "Write a response to a customer as a service rep at Saks for when they ask for a price adjustment." Now I want to show you how we can use that data that we've just connected into the platform to make this prompt much more impactful. So I'm going to go ahead and bring all of that data that we've connected from my Snowflake instance, all of that order management data and plug that here into the prompt. So for example, for my customer name, I'm going to go ahead and click on the contact record and bring in the contact name. And for the order item as well, this is where I'm going to pull in the record snapshot for the order item, which will bring in data related to the order that the customer has bought, for example, the date purchased and the price because we need these data points to determine if the item can be price adjusted. And just like that, within a few clicks, we've been able to ground this prompt with the customers' data points and give it context for a really good result. So why don't we go ahead and test the prompt that I've just built out over here with one of our contacts. So clicking on preview, down on the bottom left-hand side, you're going to see this resolved prompt here, which is the prompt plus the data that is referenced. And on the right-hand side here is the response being generated. So here we see that the response has reflected the amount that the customer can have refunded, which is great. But at the end of the day, this prompt isn't really resolving the customer question because it's still unable to determine if her item qualifies for a pricing adjustment. And this is because the prompt at this point in time does not have the search capabilities to reference the policy documentation on pricing adjustments and be able to reference and grab the data from it. Now to fix this problem and to make this prompt even more powerful, this is where we're going to be leveraging the power of Retrieval Augmented Generation, or RAG, for short. So RAG essentially allows you to automatically embed your data directly into your prompt. And when we say data, we mean all data, and that includes unstructured data like PDFs, call transcripts or in our case, a policy documentation on the price adjustments. Isn't that amazing? This is really allowing us to tap into a critical useful knowledge that today is trapped in all of this unstructured data to be able to deliver more relevant and personalized responses. All right. So now assuming that I've uploaded the policy documentation into the platform, let's go ahead and add a search retriever into our prompt over here. Now what this is essentially doing is that it's retrieving the relevant data from all the data they've connected into the platform and it grounds your prompt within. And of course, this is where we can also apply some search parameters to this prompt as well. Now again, the whole idea here is that we're making the LLM do what we wanted to do, to deliver the best customer service. All right. So now let's click on Preview again, and let's see how we did here. Now as you can see, we've got a much better response here and the response here is actually informing the customer whether or not the item is eligible for a pricing adjustment. And in fact, it's even outlining and taking action on the refund process as well. And if you move on to the left-hand side into the resolution text here, you can actually see the exact part by references the price adjustment policy documentation to determine if the purchase is eligible. Amazing. So now let's jump back on to my phone to see how all of this translates into a more efficient customer-agent experience. So let's go ahead and ask the Saks agent here about my order. And the agent first response back with the order that I'm referencing, which is spot on. As you can see, all of that order management data that we've connected from Snowflake is being pulled in over here. Now that we've confirmed that this is my order, the agent is able to resolve my question and process my refund in a matter of seconds. This is all powered by the prompt that we've just built together in Prompt Builder. Now what would have taken a service rep possibly a few hours to go through all of this policy documentations to resolve this case, our Saks agent here has done it in a matter of seconds. And we also saw here how much more powerful agents can be with the power of Data Cloud, being able to connect and leverage relevant data sources to enhance the agent experience, and allowing it to also take much more powerful and complex actions with your customers. Now this is how Data Cloud can power your Agentforce to deliver customer success for your business. Now with that, let's look into how our actual customers are getting started with Data Cloud and Agentforce. And for that, I'm going to pass it on to my colleague, David. Over to you, David.

Wonderful. Thank you so much, Reenu, for that amazing introduction to Agentforce and Data Cloud and for giving us such a great demo of the platform as well, really insightful. Now -- hello, everyone. My name is David Coghill, and I work for Salesforce leading our Solution Engineering team for Data Cloud covering the Asia Pacific region. And I'm joined today by one of our customers, Allegis Group, and we've got Saksham. Hello, Saksham. Welcome.

Hello, David. Nice to be here.

Great to have you. Now today, we're going to be walking through 5 tips for how to get started with Agentforce and Data Cloud. Now at a very high level, these tips are really simple. First tip is making sure that the organization is aligned. You've got your whole company ready to drive this project together. Number 2 is that you're picking a use case. So you can really get some value upfront from working with Agentforce and Data Cloud. Number three, identifying success metrics. So you're not just running a science project for the sake of it. You're actually going to have some great outcomes. Number 4 is making sure that you're building an architecture for a solution that's actually going to drive that success. And number 5 is making sure that you've got a road map. This isn't just a one and done and finished with your Agentic AI journey, you actually want to make sure that you're set up for success in the future. Now before we jump into the use cases, Saksham, I'd love to hear a bit more about yourself and your role at Allegis Group.

Yes. So my name is Saksham. I work in Allegis Group. I work as a Data Scientist, specialized in AI. So my role is to develop AI products, understanding the business, catering to their need. And if we talk about Allegis Group, so we lead in workforce and business solutions. So we are leading in creating solutions for the recruitment and sales.

Yes. Very cool. And so it's really about -- in terms of the role that you do, it's making sure that the experience for recruiters is really good, but then also the experience for candidates is really good.

Yes. Absolutely.

Very cool. And I know we're about to jump into the 5 tips, but I'm really excited about the work that you've done at Allegis Group. So thank you very much for being a great customer of Data Cloud and Agentforce.

Thank you so much, David.

Wonderful. Let's take a look at this first tip. And so this first tip is really about organizational alignment. Now we think, okay, this is a technology project. Why is organizational alignment so important? But really, it's because management is so incredibly complex. Now if we think about the amount of data that exists out there in the world, we know that there is more than 100 zettabytes of information that companies across the globe are storing in the cloud. Now if you're not sure about what a zettabyte is, don't worry, I wasn't either. But a zettabyte is 1 trillion gigabytes or 1 billion terabytes. So it is a massive amount of information. And if we think about what this means at the company level, this is a really, really complex problem to solve because we've got companies that have huge amounts of data, how are we going to make the most of it in a Data Cloud and Agentforce project. So the most important thing really is we have that organizational alignment. Now organizational alignment means that we need to start at the very top with the right direction for the organization. So what that means is executive sponsorship. We need to make sure that people throughout the organization know what it is that this project is going to drive and that there's someone at the executive level who is making sure that things are moving forward. Now not only do we need to have that top-down approach working really well, but we also need to break through those silos at the organizational level. So what we typically see in successful organizations is making sure that there's a Center of Excellence, or COE, who really owns this project from a technical perspective. And that's something that usually there are some great Salesforce admins there, and they are able to carry the project through the execution. Next, we want to be working through those parts of the business that are going to own the outcomes here. And we can see in this example we might have either the data teams, the digital teams and marketing teams who are really responsible for making sure that the outcomes we're trying to achieve are going to be met by this project. And then third, we want to make sure that IT and the data management teams are there along for the ride because typically that's where access to all these great tools and access to all these great information across the organization resides. So Saksham, I'd love to hear what's this process being like of getting organizational alignment inside Allegis Group?

So basically, being a data scientist, the first thing that we do have is a separate lab team. We meet with the business leaders very frequently, and we try to understand what are the business challenges and what are their needs, right? We host interactive sessions where they can actually voice their needs. And from our side, we can propose a unified data solution, including Salesforce, Data Cloud and Agentforce and address their pain points. Now what we actually need to understand is what would be the higher priority cases, right? Once we have the higher priority cases, we can actually develop a very simple MVP to demonstrate the capabilities of Data Cloud and Agentforce together. Now this hands-on approach actually helps them to visualize the benefits and in return, they can have a confidence when they actually visualize a product in front of them and see how can they actually meet their requirements.

Very cool. And I think that's probably something that's been a really big part of the success that you've had in Allegis Group, and why you've been able to move so quickly with everything as well is having all those layers of the organization aligned and ready to go. Very cool, very cool. Let's have a look at tip #2, and that is making sure that we're really picking a use case. And when it comes to picking a use case, it's not just a matter of saying, hey, out of the blue, what is it that we want to do? But it's about being quite deliberate and quite thoughtful. And what we see is that there is a good way of going about picking a use case, but there's also a great way of going and picking a use case. So the good way is to say, "Okay, what's the technology that's out there on the market? What are the things that we can pair technology up with people to do? And then what are the outcomes that we could achieve once we work through that?" Now what we find with our customers is that -- that is a good way to get outcomes. But if we really want to achieve some great outcomes, that's where we need to start thinking with the Indian mind. And what that means is thinking about the outcome that we want to achieve then looking at what's the technology that can help us get there, what are the processes that we can help get there, and what are the people that we need to put in place to really make this technology a success. So I want to take a look at an example of what those use cases could look like. And on this next slide, what we see is a 2-pane example. On the left-hand side, we've got what are the use cases the business wants to achieve. Now this is a fun example. This is a theme park that's working with Salesforce. Now personally, I love theme parks. I love recruiting companies as well, but I do love theme parks. And we can see some examples here where we want to enable the cast members who are customer-facing, so they're in the park, actually interacting with customers. We want them to be able to have access to data at their fingertips through an Agentic user interface. But we also want customer service representatives who might be behind a contact center, working in a contact center, we want them to have that great Agentic experience as well. You can see on this slide that we've got some great examples of what it is that we want those people to be able to achieve. On the right-hand side, this is where we start talking about the capabilities and the technology that we need in order to meet those use cases. And so it's really important that we're looking at the use case and the outcome first, then we're looking at the solution capability next. So Saksham, how have you achieved that in Allegis Group?

So basically, if we look at our high-impact use cases the first thing that we definitely look at is the business value. And the next thing is what difficulty are we going to face while implementing it? Now when we identify use cases, like I said, it has to be providing us high business values or of high-impact areas. But looking at the lab team, we assess the technical feasibility of each use case that we have of our existing technical stack. Also, we ensure that the right data is available. So the data quality always has to be top-notch when we are tackling a business problem. Like -- I mean, if you do not have the right data, there's nowhere that you're actually going for. And I always go with that particular moto, especially in AI, the amount of garbage that you put in is the amount of garbage that you're going to get out of AI. So it's really important to ensure that there's a proper data quality being maintained when you look at these high business value or high business use cases that we have. Now one such use case that actually comes in my mind is matching candidates to opportunities based on their interest, based on their skills. But also if we literally have to maintain an equilibrium, it's very important to ensure that the hiring managers or the clients that we have get the right candidates.

And I remember the very first time I heard about Allegis Group embarking on this use case a few months ago, I thought, Wow! That is a really interesting use of how you're putting the business data that you've got there as well as the business requirements that you've got to really create something that you couldn't have done without this sort of technology before.

Exactly.

And I love that you started with the business value because if we take a look at our next tip, Tip #3, it's really about identifying success metrics. And that's really saying what do we hear that we're trying to achieve? And so we can see we're back on that same use case again. But now on the right-hand side, we've got the business value that we hope that we're going to achieve with these different use cases. Now business metrics are going to change based on the use case and based on every organization, you can see some examples here where we want, for example, customer value to go up, we want revenue per customer to go up, we want customer satisfaction to go up and we also want things to be more effective and more efficient. If we take a look at the next slide, we can see even more examples of what that can look like, where we have some customers who are really going to focus on data integration and quality. Now that's really important for some of our customers with their first use case. What we also see, and this is often the real focus is business KPIs. What are the impacts on revenue and customer support going to be? We also get things like marketing metrics, how do we make sure that the engagement of our customers is going to go up once we've implemented this use case? We might also have things where we look at engagement and adoption. We might be having a look at operational efficiency or this might even be a compliance and security concern because we're now delivering something that we couldn't have done before. And now you've already talked a little bit about business value, Saksham, but I'd love to hear what was that journey like in Allegis Group?

So basically, I mean, if you're looking at the success criteria, and very rightly said that every business use case would have a different success metric. Now the first thing that we actually look at is the customer satisfaction. We measure the feedback that we get from our customers based on the quality of the candidates that they have hired ensuring that the solution actually meets their expectation. Second one is the candidate alignment. As much as importance we give to the customer, it's very important to understand that candidates are equally important in this particular business, right? We track how well the opportunities that are present to the candidates align with the skills, with the interest, balancing what are the requirements by the client and ensuring that when they are actually placed, they are happy enough and there's a lot of positivity. Now if you look at what Allegis Group has to gain, we are looking at productivity gains. We evaluate the time being saved by the recruiters on the mundane administrative tasks. We're talking about data quality. How does data quality come into the picture? You engage with the customer, you engage with the candidates, you focus on building and maintaining that particular relationship, you acquire more and more information and thus enriching the data quality. And what we always look at is the cost, cost efficiency.

That's important.

Now we analyze the cost of the recruiters versus the cost of the AI agents, but emphasizing on how much time has been saved -- how much time that has been saved translates into tangible financial benefit.

That's really wonderful the way that you are improving the efficiency of your recruiters, you're also improving the employee experience for the recruiters and you're improving the candidate experience. It's like the ability to bring all 3 of those metrics together with just 1 use case is really, really incredible. Now so far, we've talked about 3 use cases, 3 top tips, and you might have noticed, we still haven't talked about technology. So the fourth tip that we're going to bring up is actually when we start talking about what the technology is. So this is where we want to build an integrated capability and a real solution framework. So let's take a look at what this can look like. Now this slide might be a little bit daunting, but this is the full capability map of everything that's available inside Data Cloud right now. And you can see there are some that have a little pill because they're either in restricted general availability or they might be in a beta right now, but you can see that there are dozens of capabilities. Now the reason why we don't say start with technology, we say, start with a use case, is because you probably only need a certain number of these capabilities to be turned on in order to address that first use case that you want to do. And you can see on here we've got a few of those use cases lit up in blue. So we've got that we're bringing in some data. We're connecting our data sources. We're also bringing in some unstructured data. We're putting that into our Vector database so that, that can be embedded and used for the RAG pipeline in Data Cloud. We're also transforming that data. We're mapping it to the Salesforce model. So when the recruiters are logged into Salesforce, they've got all that access to data. And then we're also unlocking all of those capabilities in the activation stage so that our agents and our prompt templates are able to plug into this data that we've got inside Data Cloud as well. So the key is not to turn on everything, the key is to turn on the path that makes sense for you. And what we'll see in this very next slide is a sample architecture that we see with many of our customers. Now here's an example where we've got a customer that's got a very mature data layer, and this is at the bottom of the diagram, where they've got a whole lot of their enterprise data is already flowing into Snowflake. And this is a great pattern that we see with our customers. They've also got Amazon Kinesis that they're using to send streaming data around the organization. Now what we're able to do is take that data that's in Snowflake, take those data streams that are happening in Kinesis and make them available inside Data Cloud. We're then able to harmonize that data and then make that data available into the different parts of Salesforce that's going to help you address this use case. Now here we can see that we've got Service Cloud. This is where we're managing all of those cases, all of those customer contacts and accounts so that our customer service reps can actually interact with our customers. But then you can also see that we've got Agentforce plugged in. And this is where Agentforce is able to access all that goodness that's sitting there inside Service Cloud and via Data Cloud, all that goodness is sitting there inside the enterprise data layer as well. So we're really getting absolutely everything that Agentforce needs to be successful is being brought in through Data Cloud, made available to Agentforce. So this is what we see broadly with a whole lot of our customers, and that's tip 4, for getting started. But Saksham, I'd love to hear from yourself, what's this architecture process being like at Allegis Group?

Yes. So the very first thing is the data quality, right? You want to recognize those data sources first. Now you know what data needs to be screened. So identify those relevant data sources, assess the data quality and you start identifying opportunities for all the data to be unified within the Data Cloud. Next is data integration. Now you have data coming from various data sources, right? It could be AWS, it could be residing within your Salesforce org. Now what are you actually looking at is how can they ingest it, how can they be harmonized, leveraged within the Data Cloud, and you still want to build a very robust solution. The very next point that I can think of is the integration, right? You have identified the key integration points of how Data Cloud and Agentforce can actually come together in Salesforce as a platform, right? And how do you see the data being flowed? Now if you look at Agentforce, now Agentforce actually offers you out-of-box actions, which are already preconfigured. But to literally empower Agentforce, you can actually build custom actions. What I literally call as the 3 backbone of custom actions. One is the prompt template. Now you have Data Cloud, you have built a RAG solution, for example. Now you want to leverage that RAG, right? You can leverage it within the prompt. You can provide the context of a customized responses. The second one I would go towards auto launched flows. Now with the auto launch flows, what you can actually do is you can enable real-time data access for your agents. And the third is Apex. Now handling complex data transformation within your business logic, Apex can do it easily for you. Now just imagine an agent powered with these 3 backbones. You customize it as per your own needs and you have a very powerful agent fulfilling your requirement.

That's pretty incredible.

Absolutely. Absolutely. And the next thing that I can think of is process optimization. Now you have your mapped existing processes to identify the task you want to streamline it using AI, improving the efficiency and the outcomes. Now if we literally talk about this -- you want to leverage the automation aspect of the Salesforce and how can you actually get deeper insights of the data, which is already residing within Salesforce. Now for example, I can easily take in resume or CV of the candidate. It's more than what actually meets the eye. You really want to understand the candidate a lot better. It helps you to find better opportunities for the candidate. It also helps the client who is looking for a specific requirement. So it's very, very important where you ensure that the user experience of using that particular agent is enhanced by over these custom build actions.

It's pretty incredible. And I'm very grateful to Allegis Group for having been a Salesforce customer for many years and having all of these foundations built in Salesforce. So it sounds like you could really just start leveraging them very quickly with Agentforce. But it also sounds like you haven't been a user of the Salesforce platform for that many years, and yet you're already really deep with how all this stuff works.

Yes, absolutely. So the power that Salesforce actually has its nearly a low-code platform that makes it easier for you to create MVPs, relatively easy, I would say, for sure. And it became a lot easy for me to think of an idea and picturize it, visualize it and literally bring it on the table that here is an MVP of product that we could actually visualize and put it, and it did not even take probably 7 days to create that MVP. Such a powerful platform Salesforce is, I'm really grateful to it.

Very cool, very cool. Speaking of MVPs and going past MVPs, let's have a look at tip #5 because tip #5 is about building a road map and growing adoption. And the reason why I wanted that link from the MVP to tip #5 is that getting that first use case is absolutely great. It is absolutely pivotal for starting your success using Data Cloud and Agentforce, but it's super important to think big and start small. So start small with that MVP, but don't forget the thinking being idea of making sure that while you're embarking on setting up Data Cloud and Agentforce, you've really got a road map in mind of what it is that you're going to be building out. Now Saksham, I'd love to hear what's the road map being like at Allegis Group, and what's the future holding?

So I think our immediate focus is basically productionizing the agents, have an agent-centric solution or what we call is we would love to build the swarm of agents within Salesforce, which would actually help us in the decision-making. Secondly, we have a specialized lab team, right? We solely focus on developing, delivering and deploying agents. We are also collaborating with CloudWorks. Now they have -- I mean, they have deep expertise, I would say, in Data Cloud and Agentforce, ensuring that all our solutions that are being built are scalable. Now over the next few years, the road map that we have is we want to strengthen the connection between the clients and the candidates. We enable smarter matching and a career growth for our candidates. We would love to automate CRM tasks. You want to free up that particular time that the candidates -- sorry, the recruiters actually have and use AI to gain insights into client needs and candidate carrier operations.

Really great. And I feel like if you had started trying to do all of this as one project, it would have just been too much to bite off and too much to get started. But also, it sounds like you're being very mindful of the fact that this technology is moving really quickly.

Absolutely.

Even if we think about Agentforce, it wasn't here a year ago and yet building out that road map really important.

Definitely.

That's wonderful. So wrapping up those 5 tips. We've heard some really great insights from Saksham, and thank you for those insights, Saksham. But just to remind us all of those 5 tips: Number one, get that organizational alignment, absolutely important to make sure you've got the top-down support, but also you've got the right teams in place. Number two, make sure that you've got a strong use case that you're leading with. Pick that use case, don't start just with the technology. Number three, make sure you've got success metrics identified for that use case. So you know what it is that you're trying to achieve by doing this project. Number four, get that technology right. And yes, that is all the way in #4, don't start with the technology and the solution, that's step #4. And then step #5, make sure that you've got a road map ready to go. So thank you so much for joining us on this journey to walk through 5 tips. Now in terms of next steps for everyone in the audience watching along. Very first one, now this is a very new one, is to become a part of our Agentblazer Community. If you go to trailhead.com, you can become an Agentblazer Champion. This is an incredible way of getting started with Agentforce and includes hands-on exercises that help you really become an expert very quickly. Number two, if you are an existing Salesforce customer, you're entitled to Salesforce Foundations. That means that you can try Data Cloud for free. If you're not sure how to get started, please work with your account team and we can get you up and running. Number three, if you're not sure you want to get beyond just the starting point and you'd like to work with Salesforce's professional services team, we do have a starter bundle available to work with our ProServ team and they'd love to work with you to get up and running. And then number four, we've got a great white paper on building an AI-powered business and the 6 key questions for executives to be asking as you embark on this AI-powered journey. So there are some really key next steps. But before we wrap up the webinar, I'd really like to open it up for questions from the audience. So we're going to open it up for a few minutes now, and we'd love to see if we can get some questions flowing in. Anything that we've covered today, anything about Agentforce, anything about Data Cloud, they're all absolutely open for us to be talking about.

Okay. Now we've got our first question in, and I've got to say, Saksham, this is one that I've heard many, many times from our customers. Let me read it aloud. I'm a bit confused about the relationship between Agentforce and Data Cloud. If I want to get started with agents, why do I need Data Cloud? Now if I had a dollar for every time I've heard this question, it would be a very rich man. Great question. And I really want to answer it in 2 ways. First is really to say that Agentforce, Data Cloud, Salesforce, it really is 1 unified platform. It is a single code base. Now what that means is Agentforce and Data Cloud are actually really intimately linked. So when you turn on Agentforce, if Data Cloud isn't on yet in your organization, in your org, then Data Cloud will also be turned on. So at that level, the 2 products, the 2 parts of the platform really just work together. At a second level though, what Data Cloud enables you to do is plug into so much more data from across your enterprise. So it's not just the data that's sitting in your CRM, it's also all of the information that you've got sitting in something like Snowflake, something like Databricks, something like Google BigQuery, it enables you to make that hugely rich information available for Agentforce. So that those agents can just be so much smarter and take so much better action. I'd love your view on that one, Saksham.

Yes, definitely. What I literally want to bring out over here is rightly said, you have a unified data solution within Data Cloud. Now what Data Cloud also brings in, which can actually power up the agents is the power of native machine learning algorithms, the predictive analytics, who wouldn't want to achieve that, right? And secondly is the power of RAG. Now, it's really important -- you have a lot of information within Data Cloud or within your Salesforce or you're streaming it live from AWS, you want to have answers based on all these data sources, which are already unified. The solution that I would definitely go to is RAG, which resides within Data Cloud, right? Now if you literally look at the user experience, now just one statement in agent, you have the entire RAG-based solution powered by the Prompt, maybe the Flow, maybe Apex, all these customized actions coming, doing the job for you, right? And this is where I feel that you can actually have the entire data being flowed from Data Cloud into the Salesforce org being called into these custom actions and you have a very, very powerful solution just by typing a sentence.

Really incredible, really incredible. Let's go to the next question that we've got. And now this one, it looks like it's a comment on the demo, and I'll read it aloud. I noticed that the prompts looked very complicated. Do we need to write all that? Great question. And again, I'll answer it in a couple of parts. But the first part is, the prompts you get in, you get from them, what you put into them. So prompts are really great way, they're the way of interacting with LLMs. And it's very important for you to be very specific with what you want to get from the LLM. Now you can start simple, but that also might mean that you get simple outputs. But a couple of key provisos to this. So the first proviso is Agentforce has some really strong out-of-the-box templates for prompts. So we definitely don't want you to be thinking that you turn this product on and it's a blank slate and you've got to work it out from there. So that's a really important starting point. The second starting point is that there's a real difference between what someone who is interacting with the agent needs to type in as a prompt versus what someone who's configuring the agent needs to do when they're setting this up. And this is something that in Salesforce, we really split that because we see that we've got the admin persona, and that's the person who's setting up these prompts; and then we've also got the user persona, the person who is going and being the user of the agents. And by making sure that the admin is setting up these prompts really well, it actually means that the user is going to have a great experience and they don't need to do any prompt engineering in order to have those great outcomes. So you might find that for the admin yes, prompt engineering, it's a really important skill and Data Cloud and Agentforce are there to help you get skilled up really quickly. But for the user, it's actually a really simple and streamlined user experience.

Absolutely. I totally agree with that point. Just -- what makes it literally easier that it can be iterative process actually, you can literally start small -- please start small, that's what I always tell. Please start small. Build it up. It's important for your prompt to get aligned with the business requirement. So it's important you visualize what you want your response to be from the LLM and that's how you actually start driving the prompt. You construct it in such a manner that you actually, at the end, you get the response that you're looking at.

Wonderful. Wonderful. Now we've had another question come in. And Saksham, I think this is one for you. Let me read the question out. So it says we've considered just starting from a large language model and then writing code on top of that to achieve something like an agent, is that something that Allegis did? Or how did you get started working just with an LLM versus deciding to work with Agentforce?

So basically, now we really need to understand what Agentforce brings in and what LLM brings in, right? Agentforce is large action models. I'm literally going to put a lot of weight on actions. Agentforce is built in such a way that it can perform actions based on your queries or the prompts that you actually write in the window. So -- and if we look at the LLM, the large language model, it can understand a simple human language, a simple human instruction. Now you can use the power of LLM to create a prompt. But using that same prompt as an action, and if you can customize it as a custom action within the Agentforce, you just give it a simple statement, it can perform that particular action calling that specific prompt. And yes, exactly, that's what it will actually do.

Wow, yes. And so you've already got the actions there ready to go, and that Atlas Reasoning Engine is really powering the whole thing.

Exactly. So I feel the biggest plus point that I actually see is how does Agentforce resonate with the human brain. Just going a bit deep into the System 1 thinking process and the System 2 thinking process. Now if I say 1 plus 1, automatically you know it's addition. But that System 1 thinking, which you know you have to add, but System 2 now literally starts calculating. So that calculation is System 2 process where you take a bit more time than System 1 and then you come to the outcome, exactly the way Agentforce actually works. It performs System 1 action of identifying what specific custom action to call or any inbuilt action to call and then the System 2 performing that particular action to derive that particular response.

Amazing. And I love that we've gone from a question about LLMs to theory of mind, but I think that's absolutely amazing. Now I'm getting some looks from our producers. We've got a few more questions coming in, but I'll give them some looks back. Do we have time to take one more or do we need to wrap things up? Okay. Now if you've asked your questions, we'll leave the questions at that point. If you've still got a question, we will hand that on to your account team, so that we make sure we're getting back to you. In the meantime, before we do wrap up, I have a special thank you for Saksham. And Saksham, this is for Allegis Group and you personally for being such a great customer and for being an Agentblazer.

Thank you so much, David. Thank you for having me on this stage.

Wonderful. Thank you so much, and thank you, everyone, for tuning into the webinar. Have a great day.

Bye-bye. Thank you.