Hi, everyone, and welcome to the partner power hour for Coveo for Agentforce two point o. My name is Martin Thus. I'm a senior two product enablement manager. Joining me today is Emma, a sales enablement manager. Let's start with the standard disclaimer just to keep everything properly framed. Here are the topics we'll cover today. Emma will start us off with a brief overview of Coveo before handing it back to me for the rest of the session. I'll begin with an introduction to the passage retrieval API and case classification API agent actions available in Coveo for Agent Force. From there, we'll take a look at what's new in Coveo for Agentforce two point o and why those updates matter. After that, we'll dive into a walk through of the Coveo and Salesforce organization prerequisites and a comprehensive review of the configuration and deployment of three use cases, each paired with a quick demo. Once the walk through and demos are complete, I'll provide a high level recap of the Coveo and Salesforce organization prerequisites, touch on a few known limitations, and share additional resources like documentation and training. We'll wrap up with a brief q and a to cover any remaining questions. If you have questions at any point, please submit them via the webinar q and a function. Thanks. With that, I'll pass things over to Emma to get us started with an introduction to Coveo. Thank you, Martin. I'll set us up with an overview of Coveo, as Martin said, and our role with sales force for anyone on the call who may not be fully familiar with us. Coveo works with some of the world's largest enterprises across millions of interactions in websites, mobile apps, service consoles, agent desktops, Internets, commerce, and more. Across all these enterprises experiences, enterprises face the same challenge. Knowledge is everywhere, and users expect immediate relevant answers. Our mission is simple, to deliver AI powered relevance at every point of experience backed by secure, unified enterprise retrieval. The truth is AI can't deliver value without the right knowledge. Gartner now affirms it. AI success starts with a search and retrieval layer. As assistants and agents take action, search isn't a sidecar. It powers the experience. That's why more than two hundred Salesforce customers already use Coveo as this layer. In addition to those hundreds of joint enterprise customers, we have more than eleven years of co innovation with Salesforce across service, experience, and commerce. And now we've extended that same proven foundation into Agentforce. Salesforce actually recommends Coveo when content is complex and audiences are diverse. If you're indexing millions of items, multiple languages, strict permissions, and varied personas, Coveo plus Salesforce delivers outsized value. We don't replace Salesforce. We amplify it for complex enterprises. This relevance translates to results, higher self-service, stronger case deflection, and lower cost to serve. The savings speak for themselves here. And the bottom line is that grounding AI in the right knowledge produces measurable business impact. Now to Agentforce. Coveo supplies the content and data layer with secure connectors and a unified index in addition to multistage Rag. This means grounded answers, smarted triage and classifications, and agent prompts that reflect product truth, policies, and entitlements. Better knowledge means smarter AI agents. Coveo seamlessly integrates with Agentforce and grounds with relevant knowledge from any content source and then powers Agentforce with better, high quality answers. And the best part to all of this is that this isn't limited to just Agentforce. The same Coveo AI relevance platform that powers retrieval and relevance for Agentforce can also augment other AI agents like Copilot or Jewel. So wherever your AI strategy goes next, Coveo is built to scale with it. With that, Coveo for Agentforce is available today on Salesforce Agent Exchange, and we're excited for you to try it and show you a little bit about what that looks like today. With that foundation set, I'll hand this back to Martin to walk through the Coveo agent actions available in Coveo for Agentforce now. Thanks, Emma. Let's take a quick look at the agent actions available in Coveo for Agentforce. The passage retrieval API action is optimized for question answering use cases. It returns contextual pieces of information, also known as passages, from items indexed in your Coveo organization. This action makes calls to the passage retrieval API, which consumes passage queries per month or PQPM. The case classification API action uses natural language processing to analyze the case details, typically the subject and description, and suggest the most relevant case classifications directly in the Agentforce chat window. This action can even be used with the out of the box update record agent action from Salesforce to apply the suggestions directly to the case. The case classification API action makes calls to the case assist classify endpoint of the customer service API. While there's no specific consumption associated with these calls, you must have a minimum allotment of PQPM to use this action. The latest version of the Coveo for Agentforce extension package, version two point o, was released in September. With this version, we've made a couple of improvements to the passage retrieval API action, including adding support for flex prompt templates and support for passing custom context, both static and dynamic values in a JSON object. So why does this matter? This makes it possible to implement additional use cases beyond simple question answering, including, but not necessarily limited to, case resolution and the creation of knowledge articles or emails. Today's walk through and demo will include an exploration of the case resolution use case. Before I jump into the walk through and demo, I do have a couple of quick notes. Most of the setup we'll see came directly from examples in our documentation with just a few minor adjustments. This is meant to be illustrative of how to configure these features and not a prescriptive one size fits all solution. As our documentation states, writing prompt template instructions and creating topics is an iterative process. In the webinar chat, Emma shared links to a couple of Salesforce help articles that provide best practices for building prompt templates and writing topic instructions. In the interest of time, all of the configuration and deployment steps have already been completed in the demo environment. We won't be building this from scratch. Instead, we're going to conduct a comprehensive review of the work. This lets you see the full setup and ask questions as we go. Because this is a comprehensive review, I'm going to do my best to maintain a reasonable pace, but don't worry about catching every single click. This session is being recorded so you can go back and review any part of it later. Without further ado, let's jump into the walk through and demo. I'm going to turn off my camera to give full focus to my shared screen. First, let's review the configuration of the Coveo organization. Note the Coveo organization ID in the URL. During the walk through for the case classification use case, we'll see where this ID is used. Your Coveo index must have at least one source configured and built. In my Coveo organization, I've configured a Salesforce based source named Barca support PPH, which contains cases with comments and knowledge articles. While it's definitely not recommended in real world implementations, this source uses the everyone content security option for the sake of simplicity in today's walkthrough. Moving over to the case assist page, let's look at the case assist PPH configuration. Note the case assist ID. During the walk through for the case classification use case, we'll see where this ID is used. The case classification tab shows we're using the recommended context recognition method with the case classification PPH machine learning model. There are two query pipelines applicable to the demo environment. Hosted Insight Panel PPH is a simple query pipeline associated with the hosted Insight Panel configuration integrated into Salesforce Service Cloud. We don't need to review the details of this query pipeline as it's not directly applicable to the Coveo for Agentforce use cases we're exploring today. Drilling down into the passage retrieval PPH query pipeline, we see a condition of SearchHub is passage retrieval PPH on the configuration tab. During the walk through for the question answering use case, we'll see where else this search hub from this condition is used. The machine learning tab shows the associated semantic encoder PPH and passage retrieval PPH models, which we'll look at next. While the machine learning models page lists several models, I'd like to call out the ones directly associated with our walk through. The passage retrieval PPH model is configured to use the Barca support's PPH source. The semantic encoder PPH model is also configured to use this source. And finally, the case classification PPH model is configured to use the same source. I would also like to note the fields to provide suggestions at the bottom of the page. These are the Coveo fields whose values will be suggested by the case classification use case. We'll see where and how these are used during the walk through. The last Coveo organization configuration items to be reviewed are the API keys used in today's walk through. The passage retrieval PPH API key was created using the anonymous search template. During the walk through for the question answering and case resolution use cases, we'll see where this API key is used. The case classification PPH API key was created using the anonymous case assist template. During the walk through for the case classification use case, we'll see where this API key is used. Moving on, let's review the Salesforce organization prerequisites, starting with the AppExchange package and AgentExchange extension package. The list of installed packages includes Coveo for Salesforce six point six and Coveo for Agentforce two point o. Looking at the configuration of the Coveo for Salesforce package, it's linked to my Coveo organization. Next, let's confirm the Agentforce configuration. In Einstein setup, Einstein is enabled. In the Agentforce Studio, we can see that Agentforce is enabled and the Agentforce default agent is enabled. The last Salesforce organization prerequisite we need to confirm is a permission set assignment. The prompt template manager permission set is required for any user that needs to manage prompt templates, including system administrator users. In this case, my admin user is assigned to the permission set. This completes our review of the Coveo and Salesforce organization prerequisites. Now let's review the configuration and deployment of the question answering use case of the passage retrieval API action. The Coveo Passage Retrieval API external credentials are part of the Coveo for Agentforce extension package, but the PR API principle needs to be configured. The API key authentication parameter contains the value of the passage retrieval PPH API key that I showed in the Coveo organization earlier. The permission set shown in the principal access section is required for anyone that needs to use question answering. In this case, my admin user is assigned to the permission set. The Coveo passage retrieval configuration custom metadata type is part of the Coveo for Agentforce extension package, but the metadata type record needs to be configured. Note the Coveo passage retrieval configuration name field value of passage retrieval configuration. We'll see where this value is used in a few minutes. The search hub field value of passage retrieval PPH matches the search hub condition on the query pipeline that I showed in the Coveo organization earlier. In the demo environment, the question answering use case is deployed using the recommended option, which is custom prompt template action. In the Agentforce prompt builder, we see the answer questions with Coveo flex prompt template. Reviewing each input, user input is a free text input and represents the user's question. PR API config is a free text input and represents the Coveo passage retrieval configuration custom metadata type record. We'll see where this is specified in a minute. We won't use the additional context input with the question answering use case in today's walk through. The case resolution use case that I'll be showing later will use additional context, but it will be passed via a custom Apex class. I won't cover all of the prompt details as most of them come directly from the documentation. However, I will demonstrate how the resources are inserted into the prompt. Input user input comes from insert resource, pretext, user input, and Apex Coveo Agentforce invocable passage retrieval flex comes from insert resource Apex Coveo Passage Retrieval API prompt template. This is a custom Apex class included in the Coveo for Agentforce extension package, and it serves as a placeholder for the passages and corresponding citations returned by the passage retrieval API. Let's return to Salesforce setup. The last step in deploying the question answering use case is to configure the topic. To do this, we'll open the Agentforce default agent in the Agentforce builder. The configuration of the Barca question answering topic comes from the documentation, so I won't go into details on every field. At the bottom of the topic configuration tab, I will call attention to the last instruction where we have the PR API free text input and the value of the Coveo passage retrieval configuration name field that we saw earlier. Switching to the this topic's actions tab, the answer questions with Coveo action uses the flex prompt template that we saw earlier. For each of the inputs to the flex prompt template, the instructions provide a brief description. On the output side, the prompt response is configured to show in the Agentforce conversation rendered in rich text. Now let's see the question answering use case in action. Since this use case requires no context from the case, we'll open the Agentforce chat window without viewing a case. When Agentforce is ready, we'll ask it a question. What is Barca Skipper? And press enter. With help from the passage retrieval API, Agentforce responds to the question and provides citations. Before we close Agentforce, we'll clear the conversation history. Next, we'll take a look at the configuration and deployment of the case classification use case of the case classification API action. The Coveo case classification API external credentials are part of the Coveo for Agentforce extension package, but the CS API principle needs to be configured. The API key authentication parameter contains the value of the case classification PPH API key that I showed in the Coveo organization earlier. The permission set shown in the principal access section is required for anyone that needs to use case classification. In this case, my admin user is assigned to the permission set. The Coveo case classification configuration custom metadata type is part of the Coveo for Agentforce extension package, but the metadata record needs to be configured. Note the Coveo case classification configuration name field value of case classification configuration. We'll see where this value is used in a few minutes. These are the case assist ID and Coveo organization ID that I showed in the Coveo organization earlier. Last but most certainly not least is the field names mapping. This is a JSON format formatted mapping from the name of each Coveo field used to classify cases to the field name, not the field label, of the corresponding Salesforce fields in the case object. The documentation provides more information on where to find the applicable Coveo fields depending on the case classification method used in your case assist configuration. In the demo environment, the case classification use case is deployed using the recommended option, which is a custom prompt template action. In the Agentforce prompt builder, we find the classify cases with Coveo flex prompt template. Reviewing each input, case record is an object based input that uses the case object and represents the case being viewed. Case classification configuration is a free text input and represents the Coveo case classification custom metadata type record. We'll see where this is specified in a minute. As with the question answering use case, I won't cover all of the prompt details, but I will demonstrate how the resources are inserted into the prompt. Apex Coveo Agentforce Invocable Case Classify Flex template comes from insert resource, Apex, Coveo Case Classification API prompt template. This is a custom Apex class included in the Coveo for Agentforce extension package and it serves as a placeholder for the case classification suggestions returned by the case assist classify endpoint of the customer service API. The subject, description, and case number fields for input case record come from insert resource, case record. The drop down list enables you to scroll or use type ahead to filter the list so you can locate the fields you need. Input case classification configuration comes from insert resource, pretext, case classification configuration. Let's return to Salesforce setup. The last step in deploying the case classification use case is to configure the topic. To do this, we'll open the Agentforce default agent in the Agentforce builder. The configuration of the Barca case classification topic comes from the documentation, so I won't go into details on every field. However, I will call attention to the only instruction where we see a reference to the classify cases with Coveo prompt template action, which we'll see next. Switching to the this topic's actions tab, the extract fields and values from user input and update record actions are Salesforce out of the box actions, which this topic will use in coordination with the classify cases with Coveo action that uses the flex prompt template we saw earlier. For each of the inputs to the flex prompt template, the instructions provide a brief description. The instructions for the case classification configuration input also include the value of the Coveo case classification configuration name field that we saw earlier. On the output side, the prompt response is configured to show in the Agentforce conversation rendered in rich text. Now let's see the case classification use case in action. Contrary to the question answering use case that we saw earlier, the case classification use case requires a case record to provide context for the action. Let's view case one thousand, which is about a blinking red light on a deep sea sonar system. Before we engage Agentforce, the case details show that the boat and first category fields have no value. We'll open the Agentforce window. When Agentforce is ready, we'll ask it to classify this case and press enter. With help from the case assist no. It looks like we let's try this again. This is that part of that iterative process that I noted earlier. Sometimes you have to tweak your prompts to get it to work a hundred percent of the time. I've got mine at about ninety nine percent. Here we go. So with help from the case assist classify endpoint of the customer service API, Agentforce provides case classification suggestions. Using the update record out of the box action from the topic, we can even apply Baja to boat, for example, and press enter. Agentforce prompts us to confirm the change, and seconds later, we see the boat field is now populated with Baja. Before we close, Agentforce will clear the conversation history. Finally, we'll close the view of case one thousand, though we will return to this case later when we demo our final use case. Our last use case today is the case resolution use case of the passage retrieval API action. Before we dig into the configuration and deployment, we need to review a custom Apex class created for this use case. The code for this Apex class came from the documentation with a couple of very minor adjustments. The documentation includes a detailed description of the code so you can adapt it to fit your specific use case. We won't review all of the code, but I do wanna call out line sixteen. This populates the additional context of the Coveo request with the description of the case. In a few minutes, we'll see where this Apex class is referenced. For the question answering and case classification use cases that we covered earlier, this is the point where we would configure the action. However, we already configured the named credential and custom metadata type record for the passage retrieval API action with the question answering use case, so we can proceed with creating the flex prompt template for the case resolution use case. In the Agentforce prompt builder, we find the solve cases with Coveo flex prompt template. Reviewing each input, case record is an object based input that uses the case object and represents the case being viewed. User input is a free text input and represents the user's question. P r a p I config is a free text input and represents the Coveo passage retrieval configuration custom metadata type record. We'll see where this is specified in a minute. As with the question answering use case, I won't cover all of the prompt details, but I will demonstrate how the resources are inserted into the prompt. Input user input comes from insert resource, pretext, user input, and Apex passage retrieval with record context comes from insert resource, Apex, Coveo passage retrieval API in a prompt template. This is the custom Apex class we just reviewed and serves as a placeholder for the passages and corresponding citations returned by the passage retrieval API. Let's return to Salesforce setup. The last step in deploying the case resolution use case is to configure the topic. To do this, we'll open the Agentforce agent in the Agentforce Builder. The configuration of the Barca case resolution topic comes from the documentation. So, again, I won't go into details on every field. At the bottom of the topic configuration tab, however, I will call attention to the last instruction, which references the PR API config, free text input, and the value of the Coveo passage retrieval configuration name field that we saw earlier. Switching to the this topic's actions tab, the solve cases with Coveo action uses the flex prompt template that we saw earlier. For each of the inputs to the flex prompt template, the instructions provide a brief description. On the output side, the prompt response is configured to show in the Agentforce conversation rendered in rich text. Now let's see the case resolution use case in action. Similar to the case classification use case we saw earlier, the case resolution use case also requires a case record to provide context for the action. Let's view case one thousand, which we saw earlier in the demo of the case classification use case. We'll open the Agentforce chat window. When Agentforce is ready, we'll ask it to solve this case and press enter. With help from the passage retrieval API and using the description of the case passed by the custom Apex class that we reviewed earlier, Agentforce responds with a possible solution for the case and provide citations. Before we close Agentforce for the final time today, we'll clear the conversation history and close the case. With that, we have finished the walkthrough and demo. Let's return to the slides. Let's go through a high level recap of the Coveo and Salesforce organization prerequisites and cover a few known limitations. In your Coveo organization, to implement a use case based on the passage retrieval API action, as you might expect, you must have a passage retrieval API implementation, including an API key based on the anonymous search template. For a use case based on the case classification API action, in addition to a case assist configuration that includes a case classification configuration, you need an API key based on the anonymous case assist template and a minimum allotment of passage queries per month. In your Salesforce organization, you must have an Agentforce implementation in Salesforce Service Cloud or Salesforce Experience Cloud. If you're wanting to experiment, Salesforce developer edition organizations include Agentforce. You also need to install the Coveo for Salesforce package from the Salesforce AppExchange. This package requires a license. The Coveo for Agentforce extension package available on the Salesforce agent agent exchange requires no additional license. Finally, you need to enable prompt builder in Salesforce setup. As of the current release of Coveo for Agentforce, the passage retrieval API agent action comes with a few limitations. First, the API key used to call the PR API is limited to the anonymous search template. Since calls to the PR API are not authenticated, the agent action can retrieve can retrieve passages only from public content. Regarding the passage retrieval API itself that is not specific to the PR API calls from the passage retrieval API agent action, performance is directly related to the number of documents and their average size. For example, doubling the average passages per document may reduce the queries per second rate by half. That comes directly from the Coveo documentation. PR API implementations are also limited to English content only at this time. Last but not least, before we get to the q and a, I have a couple of slides with additional resources to enhance your learning. The Coveo for Agentforce landing page on Coveo docs will lead you to the articles listed on this slide. Many of these were used to configure the demo environment that we saw today. Our LevelUp training platform includes a couple of courses related to Coveo for Agentforce, and more will be added in the future. Are there any outstanding questions from the webinar q and a, Emma or Andrean? I don't see any right now. None. Okay. If you do have any follow-up questions after today's webinar, please feel free to reach out to your partner manager for any help. And with that, thank you everyone for attending.