Identify the Agent Use Case

Learning Objectives

After completing this unit, you’ll be able to:

• List the components of an effective plan for building AI agents.
  
• Explain what makes a good autonomous AI use case.
  
• Define an autonomous AI use case for Agentforce.
  

Trailcast

If you'd like to listen to an audio recording of this module, please use the player below. When you’re finished listening to this recording, remember to come back to each unit, check out the resources, and complete the associated assessments.

Agentforce: A Quick Refresher

Agentforce lets you deploy autonomous AI agents, creating an agentic layer to the Salesforce Platform that supports your employees and customers 24/7. While all the features you’re used to are still available, agents act as a bridge between you and the full potential of those features at scale. Agentforce includes a set of tools to create AI agents and a library of customizable use cases across sales, service, marketing, and more. These agents know your business, analyze data, make decisions, converse in natural language, and independently execute multistep tasks.

If you embarked on this Trailhead badge, you’re probably interested in implementing an AI agent. To get started, let’s get acquainted with Coral Cloud Resorts and see how it uses agents to help its guests create excellent vacation experiences.

Introducing Coral Cloud

Coral Cloud Resorts is a luxury hospitality company, with hotels operating in some of the world’s most glamorous destinations. Nora Alami, the head of business technology, has been experimenting with Agentforce on Trailhead, and she’s impressed by how fast and easy it is to spin up an autonomous agent.

Nora is excited about Agentforce and eager for Coral Cloud to embrace the technology. She knows it’s important for her to get hands-on with AI agents to explore how they work, test out ideas, and see how much AI agents can do. But Nora also knows that making the most of her company’s latest digital transformation requires smart planning.

The Components of Agent Planning

And Nora is absolutely right: When planning an agent, you have to think about it from multiple angles.

• Use case definition and scope
  
• User experience
  
• Data and technical requirements
  
• Risks and guardrails
  
• Business processes and agent design
  

Sure, you can build an AI agent in a week or even a day. But as you prototype and iterate, you also must consider the work the agent does and its impact on your organization. If you don’t, you might end up scrapping the project and failing to deploy the agent at all.

Be Strategic About Your AI Agents

Planning is a critical part of deploying an AI agent, but it’s also important to take a step back and look at the big picture. Nora knows that all of the AI projects at Coral Cloud should align with the company’s overall AI strategy, which focuses on business value and responsible AI practices. But as with any AI project, Nora can rely on strategically selecting the right use cases—minimum viable product (MVP) opportunities to automate, like answering FAQs—to quickly see benefits while building out use cases for AI at Coral Cloud Resorts.

If your organization doesn’t have a strategy in place, we suggest taking the AI Strategy badge on Trailhead. You can also set up demo sessions to see the possibilities for yourself and help you explore use cases. From here on out, we assume you’re already familiar with the process of defining your AI vision, forming an AI council, establishing AI governance, identifying AI use cases, and building a roadmap.

Develop a Strategy

As Coral Cloud develops its AI strategy, it starts by reviewing its top objectives for the fiscal year to align its AI roadmap to actual business goals. Here are two of its highest-priority objectives.

• Reduce operating costs by 5%.
  
• Improve guest satisfaction scores by 15%.
  

With that in mind, Coral Cloud’s AI council decides to kick off their Agentforce initiative by exploring use cases related to customer service. AI service agents can handle a range of routine inquiries and tasks that free up service reps to focus on more complex and high-touch interactions. They also ensure that guests receive quick, accurate responses around the clock.

Find the Right Ideas

OK, now comes the fun part! It’s time to brainstorm all the cool ways Coral Cloud can use Agentforce. What kind of work can an autonomous AI service agent do in the organization? When you start getting specific about ideas for the work an AI agent can do, you’re coming up with use cases for autonomous AI.

An autonomous AI use case is an application of AI technology where an AI agent takes an action or series of actions that accomplishes a goal or job to be done on behalf of your employees, customers, or organization.

What Makes a Good Autonomous AI Use Case?

As you begin collecting ideas for use cases, you might notice that some ideas lend themselves well to autonomous AI applications, and some don’t. As you evaluate whether your use case makes sense for autonomous AI, consider these questions.

• Value: Why are you delegating this work to an AI agent? Is the agent faster or more accurate? Will it provide a better experience?
  
• Work: Can you describe the work the AI agent will do? Do you fully understand all the business processes involved in that work?
  
• Decision-making: Does the work involve decisions and steps that can be completed without direct human input or judgment? Are there well-defined policies, rules, and constraints that the AI agent can follow alone?
  
• Risk: Can the AI agent operate within the security, legal, ethical, and regulatory requirements that apply to the work?
  
• Data: Is the data capable of supporting the work the AI agent will do?
  

But it’s also important to really understand what agents can do as you decide on your answers. As you can see with Coral Cloud, often the easiest way to get started is by building an agent and seeing how its capabilities match with the goals you have for implementation.

Define the Use Cases

Now that Nora knows what makes a good use case for autonomous AI, it’s time to flesh out each of Coral Cloud’s ideas for Agentforce use cases so its AI council can eventually assess and prioritize them. Remember, this stage of the planning process is all about the goal, not the technical solution.

Identify the Jobs to Be Done

First, describe the work the AI agent will do. Many organizations use the Jobs to Be Done framework to outline the role of the agent and the tasks it will perform. Be sure to think deeply about the work and its expected outcomes. It’s an essential step in understanding how the agent can impact your organization, customers, and employees.

Here are Coral Cloud’s top four Agentforce for Service use case ideas.

Determine the Scope

After identifying the work you want the AI agent to do, the next step is figuring out the right amount of work. What’s the appropriate scope? Your minimum viable product (MVP) should be the smallest unit of work that it makes sense to deliver. And it’s vital to clearly define—then stick to—acceptance criteria for going live. Even quick wins can be hard to score if you move the goalposts. That approach allows you to validate your assumptions, demonstrate value, manage the level of risk, and develop a plan for scaling the AI solution.

As you go through the process of planning your AI agent, you might end up refining the scope. For example, if certain data requirements can’t be met or there are areas of high risk, you might decide to decrease the scope or make the AI agent less autonomous. As your organization’s AI maturity grows, the scope and level of autonomy can be increased. The ability to start with a small use case and continuously refine both the agent’s instructions and the data it uses to complete them is key. It means you can see the benefits of deployment quickly and make the most of them over time.

Now take a look at how Coral Cloud might scope its use case for reservation management. In the MVP version, the service agent can look up reservation details and resend itineraries or confirmations. All other reservation management tasks are escalated to service reps. Then, in the next two versions of the agent, its capabilities are gradually expanded.

Define the Business Value

When you fully understand the scope of the work the AI agent will perform, you can establish the business value of your use case. Be sure to set specific, measurable goals and focus on outcomes. But at the same time, don’t be afraid to dive in! For many businesses, the best way to see exactly how an agent can deliver value is to build and use one—even if it’s just a test agent. For more information about defining the business value of AI use cases, see AI Strategy.

Here’s the business value for Coral Cloud’s reservation management use case, which aligns with the company’s top objectives.

• Reduce call volume and increase case deflection.
  
• Improve the customer experience by providing faster, 24/7 support.
  

If you need help estimating the value an Agentforce project might deliver for your organization, check out the Agentforce ROI Calculator.

Evaluate Data Readiness

Next, evaluate your data readiness. For AI agents to perform optimally, they must be powered by trustworthy, high-quality data that’s relevant to the business context. So don’t get prematurely excited about a use case if the data is insufficient.

For example, one of Coral Cloud’s top use cases involves using an AI service agent to answer common customer questions. But after a cursory investigation, Nora finds that many of its knowledge articles aren’t quite ready for AI. There’s inconsistent formatting, lack of structure, and outdated and conflicting information. On top of that, there are gaps in coverage because some information isn’t in the knowledge base; it’s in a completely different system.

Nora knows that Coral Cloud has to start with a use case that its data can ‌support, so she proposes to the AI council that Coral Cloud hold off on tackling the FAQs use case until the organization can optimize its knowledge base for AI.

But AI implementation doesn’t have to be all or nothing. Nora digs a little deeper and finds that knowledge articles for Coral Cloud’s new events offerings have been updated more recently. And they offer much higher-quality data for an AI agent. She recommends testing out an FAQ about events specifically, which can engage customers while the company updates its more general articles. In the process, Coral Cloud can even use the events agent’s successes to help guide refinements to the company’s data, leading to better answers and happier customers.

Assess and Prioritize the Use Cases

Now that Coral Cloud has defined a few use cases, its AI council can assess the feasibility and impact of the projects and then incorporate them into their AI roadmap.

If your organization doesn’t have an AI council, then make sure your business and technical stakeholders evaluate, approve, and prioritize the Agentforce use cases. See AI Strategy for more information about the factors to consider when refining your backlog and which prioritization frameworks you can use.

For its first autonomous AI project, Coral Cloud implements the reservation management use case because it’s high impact, the data is ready, and the work can be scoped appropriately for an MVP. In the next unit, follow along as the organization figures out how to architect its new AI agent.

Resources

• Trailhead: Quick Start: Build Your First Agent with Agentforce
• Trailhead: AI Strategy
• Website: Salesforce AI Use Case Library