Dreamforce 2024, the annual technology conference hosted by Salesforce, placed a major emphasis on artificial intelligence (AI) and the introduction of AI-powered service agents called Agentforce.
The conference showcased the transformative potential of autonomous AI agents, designed to perform specialized tasks and provide proactive support across various platforms. With significant updates to AI-driven capabilities in Sales Cloud and Service Cloud, enhanced features in Data Cloud, and the launch of the new Agent Builder, Salesforce highlighted how AI integration is redefining business efficiency and customer engagement. The event highlighted Salesforce’s commitment to leveraging AI and making it a focal point of the future business landscape.
Below are key takeaways from Dreamforce related to Agentforce and considerations for your organization.
Introducing Agentforce
Agentforce, a groundbreaking suite of user-friendly tools designed to help businesses deploy AI agents, was unveiled during the conference. These agents can be customized to perform specific functions within various departments including sales, service, marketing, retail, IT, and more. Agents are designed to excel at performing specialized tasks and provide 24/7 support to employees and customers.
What sets Agentforce apart from other AI agent tools is its unique ability to enable agents to analyze data, perform decision making, and complete tasks in an almost human-like way, curtailing the need for constant manual input.
Organizations can quickly develop, customize, and launch their own AI agents with Agentforce. With pre-configured agents and seamless integration with other major technology providers, Agentforce goes beyond chatbot functionality to provide highly accurate and scalable support services.
Agentforce also natively integrates with Salesforce Data Cloud, bringing together customer data and metadata across systems in real-time. This integration helps Agentforce operate with your organization’s full breadth of data to elevate context and precision, enabling agents to deliver accurate and reliable results during customer interactions.
How Agentforce Agents Work
Agents operate by gathering data from multiple sources, including customer interactions, transaction histories, and social media. This data helps the tool determine the context of queries. Using machine learning models, agents analyze the data to identify patterns and use logic to make decisions, selecting the most appropriate responses based on past interactions and the current context. After determining which action to take, the agent executes the required action, whether that’s answering a query, processing a request, or escalating an issue to a human employee.
Agentforce agents continuously learn from each interaction, refining their algorithms to improve accuracy and effectiveness. The tool updates its knowledge base and uses feedback to enhance future interactions, helping agents remain effective and relevant as customer expectations and business environments evolve.
Building Blocks of Agents
In Salesforce Agentforce, topics, instructions, and actions are key components that define how agents operate.
- Topics: These represent the specific areas or tasks that an AI agent is designed to handle. For example, a topic could be “Customer Support” or “Order Processing.” Each topic is associated with a set of instructions and actions relevant to that area.
- Instructions: These are the guidelines or rules that tell the AI agent how to handle different situations within a topic. Instructions help the agent understand the context and make decisions based on the data it has. For instance, an instruction might specify how to respond to a customer’s query about a delayed shipment.
- Actions: These are the specific tasks or operations that the AI agent can perform. Actions could include sending an email, updating a record, or escalating an issue to a human agent. Each topic will have a library of actions that the agent can choose from based on the instructions and the context of the interaction.
Together, these components enable agents to provide helpful, context-aware support and perform tasks autonomously, helping to provide a seamless and efficient user experience.
Agentforce vs. Einstein Bots
Agentforce agents and Einstein Bots may seem similar, but there are several differences between them.
- Setup: Agentforce Service Agent sets up in minutes with prebuilt topics and actions. Agents harness generative AI and large language models (LLMs) to offer flexibility and conversational capabilities similar to human interactions. In contrast, setting up Einstein Bots involves manually training natural language models (NLMs) to understand customer language and design dialogs.
- Maintenance: With Agentforce Service Agent, maintaining and improving performance are reasonably straightforward. You can update a handful of out-of-the-box topics, actions, and instructions or quickly create your own as service requests increase. Enhancing performance is as simple as adding and iterating on natural language instructions. Conversely, Einstein Bots require considerable maintenance to keep dialogs and intents up to date and to train new dialogs and intents as service demands grow.
- Context & Exception Handling: Agentforce Service Agent performs exceptionally well in understanding human language, responding to follow-up questions, and handling unplanned exceptions by maintaining the context of entire conversations (expanded memory and insights). In contrast, Einstein Bots are limited as they cannot understand human language and can only respond to the most recent question without maintaining the context of the whole conversation (very limited memory).
Getting Started
This year’s Dreamforce put a lot of focus on hands-on utilization with the new Agent Builder. You can even try it yourself.
In the following Trailhead, you can learn to associate topics and actions with agents, construct custom agent actions using prompts and flows, and build a functional Agentforce Service Agent. This hands-on project can help you understand how to leverage Agentforce to automate tasks and improve customer service in your organization.
Quick Start: Build Your First Agent with Agentforce | Salesforce
About Trailhead: If you haven’t used Trailhead before, it’s Salesforce’s free online learning platform. Trailhead offers interactive, self-paced learning modules that help users develop their Salesforce knowledge and skills. Whether you’re a beginner looking to get started or an experienced professional aiming to deepen your skill set, Trailhead provides a range of resources to support your learning journey.
In addition, learn more about agents through this helpful Salesforce article.
Business Technology Services at Forvis Mazars has certified Salesforce professionals ready to assist with your Salesforce questions and projects. Connect with us today to learn more or start your next project.