Your Prescription for Healthier Connections with Agentforce

You.

It's so good to see all of you. I'm Jen Simpson, senior director for healthcare and life sciences here at Salesforce. I want to start by saying thank you, thank you to our customers, our partners, our trailblazer community here in magical New York City and on Salesforce Plus. We're so grateful that you're here today and for everything that you do day in and day out to advance healthcare and life sciences for everyone. Our industry has seen tremendous advancements with incredible progress from. We're just gonna go to the next slide. It's hard to believe that 15 years ago we were still getting medical records off of paper. And less than 10 years ago, we were realizing the full potential of personalized medicines. And last year we were talking about what we could or hope to achieve with this new technology with generative AI. And now we are in an incredible moment where we are realizing the potential and the real impact in your organizations and in our industry, we're seeing AI used for greater efficiency and personalized engagement than ever before. Using AI to speed drug development through AI powered clinical trials and so many more incredible examples. And all of this is being done with humans in the loop, with handoffs at just the right time. This is truly a remarkable moment in our industry. Every health and life sciences organization is at a pivotal place to meet this moment with AI poised to reshape our entire health experience. And with the pressure to be more agile and more efficient, do more with less in a way we've never had to do before. Now, our industry has a lot of data. In fact, medical knowledge doubles every 73 days. Now, we know that when data and workflows are disconnected, it impacts outcomes at every level, creates incomplete health information. And we know that addressing data is critical to seeing the change that we want to realize in our industry today. Now, our industry has a lot of familiar complexities. We've got siloed systems, complex processes and changing consumer expectations. What will our kids expect from their healthcare experience? But we're in a new moment now where we can make very complex things more simple, agile and connected. How do we do that? Well, with AI agent force, we can turn disparate data into connected experiences. We can take manual tasks and turn it into high value work. And we can create more nimble and collaborative teams. We're in a moment now where it's time to take control of the complexity with new tools in a way that we've never been able to before. And that's what we're going to show you some of today. With Salesforce for healthcare and life sciences. We have the technology and the teams to help you unlock access to care with real time data. You can engage patients and providers More efficiently with AgentForce and you can connect your teams and automate tasks on a single system. No matter what part of healthcare or life sciences you work in, Salesforce is the platform to connect the health ecosystem with prescriptive solutions to help us meet this AI moment together. Now, we launched provider and payer innovation back in 2016 and 2018 and now with life Sciences we are making massive investments. I'm sure you've heard, but we launched Life Sciences Cloud earlier this year and we are just getting started. Yes, let's give a round of applause for Life Sciences. Great. We have a brilliant team of life sciences experts who are bringing you end to end solutions from medical through commercial, all in a trusted platform and we'll get to see some of that today. So with that we're going to show you a great lineup. In a short amount of time you're going to see Agent Force, specifically purpose built for healthcare and life sciences. You're going to get to see live product demonstrations and so much more. So to kick us off with AgentForce, I'd like to welcome up Sean Kennedy, VP and GM of Healthcare. Sean.

Great. Thank you, Jen. So let's now talk about how we can help you become an agent first Health organization. By way of introduction, my name is Sean Kennedy, I'm our general manager for health here at Salesforce, focusing on strategy and solutions and thrilled to have the opportunity to be able to share and chat a little bit more about this with you here today. So whether you're a provider or a payer or medtech or pharma or public health organization, Salesforce has the platform that's going to connect your ecosystem through prescriptive applications and solutions that's going to help us meet this AI moment together. We're powering AI and human experiences deployed right in the flow of work. And we're going to help transform your systems of record into systems of insight. So what does that look like for you? Well, if you're processing prior authorizations, we're going to deliver predictive insights to your fingertips so you can speed time to care. We're going to arm you with agents that are going to be able to allow you to respond to more patient inquiries so you can spend more time delivering care. And with real time benefits verifications, we're going to help you avoid costly delays in treatment. Now, we all know that your AI is only as Useful and impactful as the data that powers it. So what do we do about these data challenges? Data that's locked in an application or inaccurate data. We're working to access now trusted and effective AI capabilities. And we do that with Data Cloud to help unlock health data access across every system. So how do we do that? Well, we do that by first connecting your data. That's provider data, patient data, EMR data, medication data, prescription data, marketing data and more. Then it's harmonized and unified into dynamic profiles for patients and members and providers. Then we activate it in the flow of work so that you can take action on that for your payers, your providers. It's all built on an integrated platform that allows you to build next generation health apps. Now, I've had an opportunity to talk to a lot of our customers about their AI journey. And as you can imagine, organizations and teams, they're in different places. Some folks are well down their path of adoption, others are just getting started. But wherever you are in your AI journey, Salesforce stands ready to help and work with you. So you can start by automating workflows with bots and embedded analytics so that you can offload the work of some of your staff with assistive AI embedded in the flow of work and made more accessible with conversational AI assistance, you're going to make it easier to be able to support your staff. And if you really want to accelerate efficiency, you're going to deploy autonomous agents working within guardrails for your patients and providers and members, and handing off to a human when needed. Now, wherever you are in your AI journey, actions are available as the building blocks to power these solutions. And actions can be deployed with any level of human intervention. So whether it's assisting clinical trials coordinators, they're working to find, identify suitable candidates, or working autonomously to be able to help a patient identify identify a doctor that's right for them. And I will say that I am thrilled to announce also that now Agent Force is covered by our HIPAA baa. It is HIPAA ready for your use. That's right, give it a round of applause. And I think the idea here is that you can also become an agent first health organization. So how do you do that? Well, you can start with our pre built out of the box embedded AI. And with this you can let AI in the flow of work. And this now allows you to be able to surface next best actions to display medical history summaries and then you can adapt those for your business and you can extend it for use Cases like generating coverage documents that are specific to your organization's policies. Then you can take the success and the trust you get from that. And now you deploy conversational AI that allows you now to use harmonized data for your care coordinator, to be able to prepare for a call with a patient, to be able to close a gap in care. Then you have the ability to be able to take now data that's available and use it to enable AI agents to work autonomously with, to be able to tackle a specific job to be done, like being able to help that patient find that right doctor for them or to help them schedule their annual visit. And at each step of this journey, you're supported by a platform that delivers to you development and testing and security and governance capabilities that are gonna allow you now to deploy AI confidently and at scale. Okay, so now I am thrilled to announce that we've got a lot of capability available for AI available today, but we've got a lot more coming. And in the next couple releases we're gonna be delivering to you turnkey assistance for your employees and an autonomous agent that's able to now deliver and fulfill patient services functions like being able to find a provider in your network to verify benefits, to respond to common inquiries from patients and schedule an appointment. Okay, so you've heard from me, you've seen the slides. Let's now see it in action. And with that, I'm going to turn it over to Xavier and he's going to now show you how easy it is to build an agent with agent builder. So Xavier, off to you.

Thanks, Sean. Hey everyone. I'm Xavier Jamal. I'm a principal solution engineer on our healthcare and life sciences team here at Salesforce. Let's take a look at the patient experience with AgentForce. When patients come to your website, they're most likely looking to understand their options for care. Oftentimes having to navigate multiple pages and to get the answers that they need. That can lead to a pretty frustrating experience. In this case, our patient, Lauren wants to be able to understand her benefits and make sure that she's in network for a future appointment and her options down the line for family planning and family advising programs. Instead of sifting through the website to get what she needs, Lauren can interact directly with an Agent Force. Agent Agent Force agents operate independently and autonomously without direct human supervision. With the ability to take actions on the behalf of your patient while knowing the context of the page that the patient is on here. Initially offering help scheduling an appointment instead of calling in, Lauren wants to check the status of that insurance coverage so she can then easily book an appointment right now, helping to avoid potential prior authorization or eligibility implications down the line. This will also help to decrease your patient contact center resolution times. Here we can see that the agent actually executes the benefit verification and we can see that Lauren is in network and has coverage with Makana Health Plan and has a 10% in network copay with her plan. From here Lauren quickly continues her scheduling process. In the same window, AgentForce collects additional key information as needed before taking actions like booking an appointment, confirming preferences like seeing that Lauren previously had historically female providers from within her Unified profile. Inside of data cloud, AgentForest also understands the context of a normal conversation. For example, understanding the patient has asked for a date like Friday or next Friday or that the following Monday actually works better for her. After booking and confirming her appointment, Orin wants to pivot to understand the details around the patient and family advising program within the organization. AgentForce is able to provide the appropriate responses in context of the conversation using the all new AgentForce Reasoning Engine or the brain behind AgentForce and provide answers in natural language. Responses are grounded in your trusted CRM data, your knowledge base and any other data source through retrieval augmented generation via Data Cloud all along with your defined parameters. This process ensures that your patient is satisfied with new agentic natural language experience and of course has the option to talk to a live person, which is what we're going to go and do next. Now inside of Health Cloud a human agent can pick up right where Lauren left off with the AgentForce agent through omnichannel. Inside of our Unified profile we see in the center of the screen a full history of Lauren's conversation with the agent and along the right hand side we see the same underlying technology to assist our continued interaction, surfacing operating procedures or even reply recommendations in this instance getting the assistance needed for a question about facility fees or claims management. Now let's go behind the scenes into the user friendly Agent Builder interface which was used to build both the autonomous and assistive agents for benefits verification and and scheduling. AgentForce is comprised of three simple yet robust elements, topics, instructions and actions. Topics group actions into jobs to be done. Like here we're seeing an appointment management topic that previously helped schedule the appointment for Lauren. Behind every topic are instructions that you write in natural language. These guide your agent on how it chooses which actions to run or what clarifying questions to ask while also placing guardrails. Each topic has a specific subset of actions that the agent can then call on these actions are tied to existing Salesforce foundational tooling like Apex and Flows, so you can leverage your existing investments and your existing builds. And of course all of this can be tested and checked within the Advanced Planner service in the middle of the screen displaying inputs and outputs alongside grounding and reasoning details. All this experience is meant to be a low to no code environment so that admins can get agents and iterations of those agents up and running in a matter of moments versus weeks or months to experience the fastest time to value. And now we're going to take a turn over to our life sciences team. But let's start by watching a short film first. The excitement around health care and life.

Sciences at Salesforce is really at a fever pitch.

We're at a pivotal point not just in technology but in the world.

Life Sciences Cloud is an end to end engagement platform. We're connecting the full journey of a patient or a physician. Data Cloud can not only harmonize all.

The data within Salesforce, but it can go out to your external systems like clinical trials or SAP in the back office or take web engagement data and.

Bring it all together so that there's.

A single golden record for each and every customer.

We're able to unlock now the wealth.

Of both structured data as well as unstructured data for use in both predictive and generative AI applications.

This is a new cloud at Salesforce that was born in the era of AI. So we lead with that in all of our thinking.

And to be clear, we're not looking to replace care teams, we're helping to augment them so they can spend less time doing administrative work and more time focusing on their patients.

Me personally had clinical trials in faster. My father could still be here today.

In the not so distant future we're going to cure cancer, increase people's lifespan and we're going to do it all in partnership with the top life sciences companies in the world.

CRM, AI data and trust from clinical.

And medical through commercial Salesforce is all in.

All right?

Well as you can see, we are united in our commitment to the life sciences industry. The innovation in science that you apply to help patients across the globe truly does inspire us and fuels our software and AI innovation to help the life sciences industry become more productive. My name is Frank Defesh and I have the privilege of leading Life Sciences Cloud here at Salesforce. So I want to start by really explaining what our focus is. Our focus is is about creating better experiences for your providers, ATPs and your patients. A lot of times the information that your customers need is not what they get is because your teams are operating in silos using disparate systems that disconnects your knowledge and understanding of your customer and disconnects therefore their experience with your organization, be it a patient or provider. If what they need is not often what they get, and there's often so much information that it's almost impossible for them to consume. Now the organizations that get orchestrated and personalized, that provide the right information, will start to differentiate themselves. And technology is leaping forward to make this far more possible than it ever was before. And that will truly differentiate yourself and create a better experience. So what are we really talking about? Right, we talk about the patient experience, the provider experience. If you're providing, not understanding your patients and providers does limit your ability to work with them. If you don't understand their needs, you're not able to engage with them on the right terms. In order to do this, we believe it's important to have both a provider and a patient. 360 a provider. 360 helps you truly understand the needs and preferences of your HCPs so that you can speak to them and meet them where they want in their channel with the information that they're needing at that point in time to help them with their patients. And their needs change as they learn about disease states, therapeutic classes and understand how their patients preferences are changing. Now a provider 360 a patient 360 is important to understand the whole patient and better anticipate those critical points in their healthcare journey where they need support. Now those patient needs, they change as well as they learn about their health, as they deal with their diagnosis and as they settle into a steady state in their course of treatment. But the real sweet spot that we think is going to be unlocking a lot of new capabilities is where those two intersection points meet. Where you have an engaged and educated HCP and an engaged and educated patient. Putting those two things together can really advance healthcare. Now while Life Sciences Cloud is actively being used by many organizations for med tech sales and patient and medical call centers, clinical recruiting, patient engagement, advanced therapy management, we're really excited about the innovation that we're bringing to market for direct HCP engagement. We know a lot of you have a choice to make and yes, you do have a choice. We are excited to announce Life Sciences Cloud for Customer Engagement available next September. Now while our solution will include known industry capabilities like an offline mobile application and omnichannel content and PDMA compliance, sampling and medical inquiries, none of that's new. You've had that for years. But we want to move the industry forward. So what we're doing is we're connecting those capabilities and we're ingesting that with embedded data and AI. Imagine the medical calls, the medical inquiry process, right? We are going to unify that from the medical inquiry captured on an iPad in the field to self service on the web and into the call center by leveraging robust case management part of service cloud, which is included inside a life sciences cloud. What's even more exciting is the embedded analytics and AI with assistive and autonomous AI to provide your users with real time information so they can identify the right people to talk to and make those interactions more meaningful. Yes, I am talking about Agent Force for Life Sciences, so I'm happy to talk about Agent Force for HCP engagement, the endless possibilities that we can apply here and they're very powerful from total autonomous agents to make self service as powerful and as delightful of an experience as we all would hope to assistive AI to help your commercial and medical teams prepare and plan and so to establish more meaningful interactions. Take for example a call advisor agent that streamlines collecting all the information that people have historically done manually to do their pre call planning to get ready. That will help your commercial teams engage better to an MSL agent that will allow your medical teams to respond to medical inquiries more accurately, to engage with KOLs more directly and to detect more medical insights from all the engagements across your interactions with your customers. But enough of me talking about this, let's actually show you. So I want to pass it over to my friend Joe Ferraro to talk to you about how easy it is to build with Agent Builder.

All right, Frank, Frank, Frank, thanks so much. So Frank, as you know, we've been showing a lot of demos of our mobile app experience on the iPad for Life Science Cloud for customer Engagement, which will come out in September 2025. There's a lot of work within our product organization right now focused on making sure that we hit that deadline. And so but what I want to do is peel back the curtain a little bit because we get questions where it's like, wow, this looks like magic. We have this conversational assistant that's working together with this rep. How does this all work together? And so what I want to show you is really the back end of this. How do we configure this? How do we use data cloud, how do we use Agent Force and how do we use life sciences cloud together in tandem to be able to deliver this solution because my product team is actually building on these GA solutions today when we drive towards September 2025. So to do that, I'm going to take the role of an admin here from our fictitious pharma company, Makana Pharma. So we're going to start in Data Cloud and we're going to talk about these data streams that we brought in from our Data Lake. So these are data streams that are going to power our functionality that Frank was talking about, whether it's pre call planning, whether it's next best action. And these are data streams that can come in from your existing lakes. You don't need to move away from those investments that you've already made in those data lakes. Those can be reflected inside Data Cloud and used to power those AI experiences. And speaking of next best action, I want to show this amazing machine learning model that our data science team here at Mekana Pharma has built. So they built this propensity to prescribe machine learning model and They've used AWS SageMaker to build this model. And what's so cool about this is we're using the bring your own model functionality inside of Data Cloud to actually bring this model into Salesforce and be able to leverage that in AgentForce. But what's even cooler is this model is actually getting trained by data that's coming in from Data Cloud. So imagine all of those interactions that you're collecting from those HCPs, all those calls, all those insights, all of that data that's landing in Data Cloud is actually being used to train that model in AWS. And that SageMaker model can be used to power experiences inside of Agent Force. So let's switch over to Agent Builder here. This is where we build our AI assistant. So again, we've seen the, the conversational experience on the iPad application. This is the back end. This is how we think about building these solutions. So in the Agent Builder you'll have agents, as Frank alluded to, for MSLs, for CAMs, for field reps. So we're going to focus on our sales rep agent right now. And so an agent is comprised of topics that it has access to and then the actions that are behind those topics. So we'll start with this HCP engagement topic that we have. So within this topic we can set the scope of what this particular topic can do. So, you know, basically provide an assistant, an AI assistant for a rep that's in the field. And here's some sets of instructions that we're providing this particular agent to make sure that we're dictating how this particular topic responds to prompts that are input by a rep in the field. And these are all human natural language instructions that we're providing it and then a couple of examples of what user input should be. So for example, on that next best action example we've got a prompt that would be something like who should I meet with next? Or can you summarize these HCP interactions? And then of course along with that topic we've got a series of actions, right? And we can see a couple ones that are familiar here. So maybe suggest the next best HCP engagement. So again, backed by that model and SageMaker, we'll come back to this in a minute. Add an HCP to a marketing journey. So imagine being able to just tell AgentForce or to use a pre can action to be able to onboard an HCP to a marketing journey or generate an HCP interaction summary. So summarization is something that Gen AI and LLMs are really, really good at and something that's great for pre call planning. So let's look at what that configuration looks like inside a product. So again, all of this is declarative. We're not writing any code to be able to provide this functionality to our users on our field rep. In our prompt template here we're using natural language. I think we've all done or become familiar with prompt engineering, right? And being able to use natural language to tell the LLM how it should respond. So in this particular case, the use case is we want to be able to provide a sales rep summarization of interactions that that HCP has had with my organization. So we're writing a system prompt. You're a highly specialized gen AI system and you're providing expert level insights to pharmaceutical field representatives. These are your goals. Here are the guardrails. When I ask you to generate a text message, here are the instructions that you need to follow. So you need to generate a response in this specific format. You need to make sure that you're keeping in mind these regulations and we can put this is where we have the opportunity as a customer to to be able to implement our own compliance guardrails, our own business rules, but in natural language. And of course in order to generate the summary you need the data. So backing the summary is the data, cloud data that we brought in through those data streams, any of that CRM data and we're using a flow, a no code flow to retrieve all that data and actually inject that as context into the prompt. So when we want to test what this particular prompt looks like, we use the preview button here. And what this is going to do is it's going to take all of that system prompt that we've created for this particular prompt, it's going to grab all the data, it's going to be contextualized to the account that I want to test on and to the user that would be potentially running this. And this is the process of doing that prompt engineering where you have to kind of, you have to toy with the prompt, make sure that you're telling the LLM to respond in the right way. And then of course we have the output here that we can verify before we activate our prompt for the action. The other key point here is we have a bring your own model capability here as well. So if you want to use the out of box models that Salesforce provides, you can test those for whatever model maximizes the type of output that you want. And then of course there's the opportunity for you to bring your own models as well. So switching back to the agent builder, we have our prompt written. Let's go ahead and test what this looks like. So what we have is the ability to use this conversation preview functionality to be able to input our prompts. So in this case summarize HCP interactions. Right. And we'll get to how we can make this a pre canned action in a second. But I'm typing this out and agentforce is going through and using the reasoning engine to determine, okay, which topic am I going to use, which actions are available in that topic, what is the context available to respond to this interaction, this particular prompt? And then who's the user? Should I, should the user have access to the data that they're requesting? And it's using all of that to be able to provide this output. And of course there's traceability as well. Right. So you see what the user prompt was, you see the topic that Agent Force selected, you see the actions that were executed in the response. So let's go ahead and challenge it here for a second and let's get a little bit more information about Case the. So this is a medical inquiry, Remember I'm a sales rep. Let's go ahead and see what AgentForce decides to do with that. So the response that we get is, hey, medical inquiries include restricted information which is only available to specific roles. I can't provide those details to you right now. So this is the Einstein Trust layer coming into play where it knows that I don't actually have access to detailed information about that case. And then of course, I've also augmented that with my own prompt and my own compliance guidelines and that I've injected into the prompt to say, hey, if the user wants more information about medical inquiries, just respond politely. No. So this is the back end. This is how we configure these and once we're ready, we can activate this particular agent and then we can see what this looks like on the rep side when this particular agent is made available inside the iPad. So let's go ahead and switch gears now to our iPad experience. So what we're seeing is again, Agent Force. Right? So this is the front end that we've been showing you for the last several months. And then what you've seen over the last few minutes is what that back end looks like. So let's go ahead and say, who should I meet with next? Sophia's got a few extra minutes in her day. She needs to figure out who she should meet with. And that machine learning model that we brought in from Sagemaker is providing recommendations on who she should be meeting with. So let's go ahead and meet with Dr. Jones. And so this is where that prompt that we just engineered comes into play. So I get that summarization of all the information about Dr. Jones and maybe I drill in and say, hey, give me a little bit more information about her content engagement with Immunexus. And so we get that and so let's press it again and see if we're going to be able to keep on the compliance guardrails. Give me the details of the conversation between kay Duffy, the MSL and Dr. Jones. And then once again, you can see the agent is not going to respond with that information. It's staying on the compliance guardrail. So this kind of gives you the backend configuration of what this looks like when we're building these agents inside of Agent Builder and gives you the front end experience of what that interaction is going to look like for the rep. We are super, super excited about this technology and bringing this to the market across all the Personas that we service in Life Sciences Cloud. So with that, back to Jen.

Thanks Joe, for a great demo. As we wrap up, we want to encourage you to continue your learning journey, learn more about our products, about our solutions, Check out our Einstein One editions. We have Health Cloud, we launched Life sciences Cloud, Cloud, Einstein One Edition. Check out AgentForce for healthcare and Life Sciences. Get your hands in these amazing agentforce demos and then explore the campground, talk to a product expert, talk to a health expert. And with that, we want to thank you. Thank you for joining us and enjoy the rest of your day.