AI moves pretty fast. If you don’t stop and look around once in a while, you could miss it.
With that adage in mind, it can be helpful to take stock of where we currently stand in the ongoing wave of AI transformation. We’ve broadly moved past the initial shift sparked by ChatGPT and OpenAI that oriented around generative LLM interactions and copilots. Although these solutions very much still have a part to play, the current overarching trend–what we’re referring to as the third wave of AI is agentic.
Agents are distinguished from the first iterations of copilots because they have the autonomy and functional capability to take action, much like a human would in executing a task or function. With this autonomy, Agents are becoming extensions of our teams and deployed across the enterprise.
At MuleSoft, we’ve adopted a two-pronged approach when it comes to the agentic wave.
First is our commitment to supporting Agentforce. We truly believe that APIs are the foundation of Actionable AI and with MuleSoft for Agentforce your Agentic use cases can include actions taken across every business critical system, workflow or process from your existing API investment and digital ecosystem.
For many years, integration, automation, and APIM were the foundation from which digital transformation flourished. AI represents the layering of cutting-edge technologies like RAG, vector stores, and LLMs, and these tools span beyond Agentforce. MuleSoft’s capabilities need to be at the forefront of these technologies as we were in the era of digital transformation, and our customers will need the appropriate tools to address this new reality.
From this need, we’ve seen a huge reaction from our user base, both in upskilling with Agentforce and building on their MuleSoft expertise to address the nascent challenges of AI.
One of the most prominent examples is the MuleSoft AI Chain project. This open source initiative is a joint effort between MuleSoft and our community to create a collection of connectors that make it easy to weave these new AI technologies with Anypoint Platform.
There are plenty of amazing use cases and resources you can read up on from the project, but the focus of this content is on a subset of these connectors that are now supported by MuleSoft and available today in Anypoint Exchange, the MuleSoft AI Connectors.
What are the MuleSoft AI Connectors?
There are currently three MuleSoft-supported AI Connectors available on Anypoint Exchange: the AI Chain connector, Einstein Connector, and Agentforce connector.
The MuleSoft AI Chain Connector was the first connector released as part of the open source project and also the first to be officially supported. In essence, it is the core connector of the initiative, providing a unification layer to easily swap between LLMs and chain them together, providing organizations flexibility in AI tooling.
The Einstein Connector allows MuleSoft customers to interface with the Einstein Trust Layer. As you’re feeding customer data and prompting Large Language Models (LLMs), it’s important to have a trusted, secure interfacing layer to parse sensitive data to. That’s why Salesforce released the Einstein Trust Layer and, with this AI Connector, users can interface their choice of supported, third party LLMs through Salesforce’s Models API.
Finally, we have the Agentforce Connector. The most recent of the Connectors released, it’s the first to be developed in-house by MuleSoft and contributed back to the open source project, and aims to build upon the foundation of our MuleSoft for Agentforce initiative.
Learning more about the Agentforce Connector
The main thrust of our MuleSoft for Agentforce solution approach has been centered around exposing MuleSoft assets like Integrations, automations, and applications in a secure and governed way to Agentforce. To build a foundation of actionable AI, agentic tools require access to all of your systems, processes and workflows and the iPaaS layer is perfectly situated to offer that. With the Agentforce Connector, we wanted to invert that paradigm and provide our customers to be able to invoke an Agentforce Agent from MuleSoft to call upon them to act autonomously.
This connector enables Mule application and API interactions with Agentforce, the agentic layer that sits on top of the Salesforce Customer 360 platform. This connector enables MuleSoft users to bring real-time context and data from other applications into Agentforce interactions.
A key benefit of this connector is that users can leverage data and context from previous steps in a MuleSoft integration flow. Let’s take the example of a technology company that prides itself on delivering top-notch software and hardware products to enterprise clients. When a customer reports an issue through Jira, their customer service agents are faced with a cumbersome process. They have to toggle between Jira and Salesforce, manually gathering details about the customer, their company, and their history with their products– a time-consuming and inefficient process.
To streamline this experience, an integration specialist can configure the Agentforce Connector, linking Jira with an AI agent in Salesforce. This AI agent, with access to Salesforce data and a powerful language model, automates the research process. Now, when a new issue is created in Jira, the AI agent performs all the necessary research and returns a summarized synopsis. This allows customer service agents to focus on resolving issues more efficiently, significantly enhancing their productivity and showcasing the value of the Anypoint Agentforce Connector.
This is just one of the many different use cases possible when you can invoke the power of Autonomous AI via Agentforce from your existing Mule applications.
Resources
There are many resources available via the MuleSoft AI Chain open source project, including a very active YouTube channel which highlights a huge array of different use cases. You can also access the GitHub repository for the open source connectors as well as the supported AI Connectors on Anypoint Exchange.
We’ve also recently released a four part video series that covers the MuleSoft for Agentforce experience end-to-end, including the MuleSoft AI connectors, be sure to check that out.
Stay tuned for more announcements regarding the MuleSoft AI Connectors. It’s our intention to support the most popular open source connectors, while simultaneously developing our own in-order to continuously support the MAC community.
The Vectors Connector, which provides access to a broad number of external vectors and databases, is next on the docket with the AI Connector available on Exchange later in 2025.
