Companies are focused on selling products, solutions, and services that can increasingly satisfy their customers’ needs, and most are using AI to streamline their processes. But business ‘use’ isn’t the only way these new technologies can be implemented. We’ll take a look at a more entertaining, less business-centered use case involving MuleSoft’s AI Chain and Agentforce to create an AI-powered version of the classic Hasbro game “Guess Who?”.
Understanding the MuleSoft AI Chain
The MuleSoft AI Chain (MAC) is an open-source project empowering developers to integrate advanced AI capabilities into the MuleSoft ecosystem. It is composed by assets, each representing a specific connector within the project allowing for:
- Easier interaction with LLMs and Vector Stores
- Optimized usage in MuleSoft applications
- Access to a wide range of AI services, tools, and chains
Key features of the MAC include:
- Dynamic and flexible tooling: Enabled through configuration files and Anypoint Exchange
- AI lifecycle management: Agent lifecycle management within Anypoint Platform
- Centralized AI agent design: Via the Anypoint Design Center
- AI agent monitoring: Via Anypoint Monitoring and Visualizer
- Low-code development: Supported by Anypoint Studio and Anypoint Code Builder
- Unit testing: Integrated MUnit framework for robust testing
MuleSoft AI Chain contains several connectors inside, each having different functions to meet different needs and purposes:
But for our example (game development), we utilized the Agentforce connector. Before diving into the nitty-gritty, let’s explore Agentforce in detail.
Agentforce and Mulesoft
We’ve built Agentforce, an all-new layer in the Salesforce Platform for deploying autonomous AI Agents that work alongside humans to get jobs done.
Agentforce helps build powerful, autonomous agents for sales, service, marketing, commerce, and so much more. This is a great advancement for our AI technology capabilities, because unlike copilots, Agentforce is autonomous.
Agents don’t require human conversations to take action; they can anticipate, plan, and reason with minimal help. They can automate entire workflows or processes, make decisions, and adapt to new information without human intervention. Agentforce ensures a seamless hand-off between AI agents and human employees, facilitating collaboration across every line of business.
Agentforce enables users to:
- Deploy out-of-the-box (OOTB) agents for any role or industry
- Enable agents to take action with existing workflows, code, and APIs
- Connect your agents to any enterprise data without training AI models
- Deliver accurate actions powered by the Atlas Reasoning Engine
To generate relevant responses and take action autonomously, agents need access to the external systems where information, workflows, and processes live.
This is where MuleSoft for Agentforce comes in. MuleSoft is a foundational component of Agentforce that makes it possible to connect any system so that Agentforce can securely perform actions across your entire business.
MuleSoft for Agentforce enables users to:
- Equip agents with the information necessary to deliver relevant and personalized customer and employee experiences by connecting any third-party, legacy, or hard-to-reach system.
- Drive efficiency and innovation by enabling Agentforce to take action autonomously, so that employees can focus on more complex and higher value tasks
- Reduce security risks by ensuring every external Agentforce interaction is managed and in compliance with company policies
A key differentiator for MuleSoft is that we help customers connect to any system, build securely, and automate workflows with one platform. MuleSoft for Agentforce empowers devs and admins to enable Agentforce to take action across any third-party system that exists within your organization.
It provides the tools to:
- Connect siloed systems and information, regardless of its location or data type. Modern applications, legacy systems, unstructured data and more come together with the platform’s core capabilities.
- Secure every system you bring into your enterprise with universal API management and governance to ensure data breaches and security risks are eliminated
- Action across any third party system or application with new tools that allow developers to expose APIs as Agentforce Topics and empower admins to build workflows across popular external systems with OOTB connectors
The game-changer: MuleSoft, Agentforce, and AI in action
We understand how MuleSoft, Agentforce, and AI power cutting-edge business solutions, but what if these same powerful technologies could also ignite innovation in unexpected, playful ways?
Imagine harnessing the sophisticated integration of MuleSoft and the autonomous intelligence of Agentforce to transform a classic game into a dynamic demonstration of modern tech capabilities. Our goal is to provide you with a glimpse into the limitless potential of these tools, proving that the ingenuity fostered in play can directly translate into groundbreaking business applications.
How we used Agentforce to reimagine our “Guess Who?”
The original “Guess Who?” game is a two-player board game in which players each guess the identity of the other’s chosen character. It contains two boards where players flip down pictures (that have names) of characters after guessing whether or not that person is the right person their opponent is thinking of.
While other versions (including digital) already exist, reimagining the game with AI technology requires special attention to correctly manage questions, answers, characters to eliminate, and how to maintain a regular game-playing pace and experience with an agent.
Examining further, the connector we used to interact with the AI is the Agentforce Connector.
The Agentforce Connector enables developers to easily integrate with AI agents in Salesforce from within Anypoint Platform. Anypoint users can connect external systems with Agentforce agents, allowing them to power new experiences, streamline previously manual processes and deliver on the promise of cutting-edge AI technologies.
To use this connector, you need to configure an Agent on Salesforce, to which you can then assign Topic and Actions.
In this case, we have an Agent with 2 Topics associated that refer to different Actions:
- One Topic is used as an utility, useful to filter the set of elements composed of characters that are to be kept or eliminated and then to reconstruct the new updated dataset with the characters to be guessed.correct.
- The other Topic is made up of actions and instructions that explain the rules of the game to the Agent and how he must interact, following very specific rules (e.g. it must answer only with ‘Yes’ or ‘No’, or it must ask specific questions about the physical characteristics of the character it’s guessing).
This is the configuration of the Topic “Guess Who?” game EN:
“Guess Who?” powered by Agentforce game flow
How does this iteration between the user and agent work? Let’s go over the game flow.
First, the user accesses the game screen, enters their name, and logs in.
After logging in, the agent greets and welcomes the user and starts the game. Once the user has chosen the character, they can start answering the first question from the agent (who will have to guess the chosen character).
After the first question and the first answer, the game continues with other questions asked by the agent, which the user will answer. As the user answers, the guessable character database reduces; characters that don’t meet certain physical characteristics are eliminated.
Once there is only one character left, the agent is ready to guess the character the user has chosen. It can also hazard a guess when there are still two to three characters left in the dataset. Once the correct character is guessed, the game reaches completion and starts again.
Here’s a demo of our “Guess Who?” game in action, with two user challenges vs. an Agentforce agent.
From work to play, MuleSoft and Agentforce do it all
Technology permeates every aspect of our lives, fueling innovation in integration, automation, and AI. The “Guess Who?” game exemplifies how the advanced solutions from MuleSoft and Agentforce, typically used for intricate business operations, also possess immense potential for ingenuity and creativity.
This playful application serves as a compelling testament to how MuleSoft’s foundational connectivity and Agentforce’s autonomous intelligence are tools without limits of time, place, or imagination, proving that the innovation sparked in fun can indeed translate into groundbreaking advancements, not only at a business level but also in how we perceive and utilize technology itself.
