As a Salesforce Admin, you’re likely already familiar with Agentforce, Salesforce’s collection of assistive and autonomous artificial intelligence (AI) agents. Agentforce can handle a wide variety of tasks within the Salesforce ecosystem. But what happens when your business relies on external applications? How can you extend Agentforce to integrate with other essential tools?
Enter MuleSoft–Salesforce’s leading integration, automation, and application programming interface (API) management platform. You might have experience with some of the capabilities of MuleSoft APIs. Whether building application integrations in Flow or expanding the scope of Data Cloud, MuleSoft is the go-to tool for extending these core capabilities beyond the reach of the Salesforce ecosystem.
Admins can now add Agentforce to that growing list of work areas that MuleSoft augments. As we transition into the agentic wave of AI, the ability to easily integrate third-party data sources and workflows is paramount. The world of AI is a world of APIs, and MuleSoft is the best-in-class tool for extending your Agentforce reach to encompass every application process your business relies on.
Building agents is a new frontier for admins, and provisioning them with access to your most critical IT assets may seem daunting. That’s why MuleSoft for Agentforce is built around three central tenets to ease any apprehension you may feel at the prospect.
MuleSoft is releasing a variety of features to enhance Agentforce in line with these three core tenets. One such feature is MuleSoft for Agentforce: Topic Center–our latest product innovation designed specifically to augment Agentforce.
To highlight the value of MuleSoft for Agentforce: Topic Center, we first need to define what a topic is.
Topics are a way to package a list of actions and instructions on how those actions should be used by Agentforce. Topics can execute actions either inside or outside of the Salesforce ecosystem.
Agentforce has a set of actions and guardrails as part of its scope. As you expand the capability of an agent to include third-party app functionality, you need to provide guidance for how it should act to achieve a desired outcome.
That’s the function of a topic—it’s a collection of paired actions and instructions that form a set of capabilities that an agent can execute across third-party systems. They ensure that as you extend your agent’s capability, you’re doing so in a measured way.
Topics can be created directly in Agent Builder, so why would you want to create them using MuleSoft?
MuleSoft developers are already building out business-critical integration infrastructure, and Topic Center weaves Agentforce Topics directly into that API lifecycle. Enabling Topics as part of the API development process is now a simple checkbox.
The MuleSoft developer building the API–the person most familiar with the capabilities, actions, and guardrails required in an Agentforce context–can generate topics in-line as part of the specification. Topics created here fall under the enterprise-grade governance and security protocols that MuleSoft provides out of the box, so you can leverage them confidently.
Something else to note: The topics created in MuleSoft can also be managed programmatically. Topics created in Agent Builder are currently tied to the agent they were created for, whereas MuleSoft Topics can be utilized across multiple use cases. MuleSoft also enhances the strategic growth of Agentforce–users can monitor how Agentforce utilizes third-party actions and develop their agentic strategy accordingly.
Thanks to MuleSoft for Agentforce: Topic Center, agents are now another channel you can extend. Agentforce’s contextual reasoning can decide how to best use an integration to fulfill a given prompt, without the need for you to get hands-on with the logic. This reduces the expertise required to leverage more technical integrations and makes integration conceptually more accessible.
Broadly speaking, this will have two major areas of implication for you as part of the admin experience.
First, MuleSoft is the best solution for you to extend your Agentforce capabilities to include apps outside of the Salesforce ecosystem. If you don’t have a MuleSoft practice in-house, it might be time to start rallying to change that. If there’s an existing MuleSoft practice, then the relationships for admins will strengthen. Admins will need to define the strategic vision for Agentforce and work with their MuleSoft counterparts to make it a reality.
The second implication has to do with workflow. When it comes to consumption, it’s grab-and-go for Salesforce Admins. You’ll have easy access to all of the third-party connections MuleSoft has to offer via the MuleSoft for Salesforce: API Catalog. You can assign these topics to Agentforce with confidence knowing that, from the initial design through to API management and the topic being posted onto the API Catalog, what they’re empowering Agentforce with has gone through all the necessary design decisions and governance to act in third-party systems effectively.
That confidence, along with the ease of adding third-party actions to Agentforce via topics, is the true value of MuleSoft for Agentforce: Topic Center, and it’s as exciting a prospect for admins as it is for MuleSoft developers.
Let’s take a basic Agentforce for Service use case as an example. Within the Salesforce ecosystem, Agentforce has a broad range of functionalities; it can execute on order management, FAQs, and much more.
Let’s say a user prompts Agentforce inquiring about an order. Drilling into order management, Agentforce can look up the order, offer alternate product recommendations, or process an exchange or return. These actions will have an associated set of instructions that provides guardrails around how Agentforce can fulfill the user request.
If there’s an order delay, Agentforce will communicate that to the user. The user may then respond that they would like to cancel the order. Based on this new context, the agent leans on its order management instructions and actions an alternative product recommendation alongside a discount to try and reduce the chance of lost revenue given the circumstance.
This is a fully self-contained Agentforce workflow that can occur entirely within the Salesforce ecosystem. There are most likely implications on third-party systems, though, like Logistics, Inventory, and ERP–tools that aren’t traditional components of Salesforce that require third-party app integrations and IT investment.
MuleSoft for Agentforce: Topic Center bridges that gap for Agentforce, providing an expanded use case that can incorporate these external systems.
This doesn’t happen by magic–it requires collaboration between the MuleSoft developer and Salesforce Admin. The admin defines the parameters of the agent to the developer and ideates on possible scenarios and solutions. The developer, in turn, communicates the capabilities of the APIs and apps available, and builds a technical solution for the admin to leverage when establishing the agent.
With that in mind, let’s go through the logic an admin and MuleSoft developer could devise in such a scenario.
If a warehouse is unable to fulfill a delivery, the logical next step is to check if there’s inventory at nearby stores. Perhaps there’s stock at a location within driving distance–is the customer willing to travel for their order? If so, then the item would have to be held for pickup. This is the ideal scenario and the option the agent should offer first, as it involves the least amount of logistical hassle for the business.
But if the customer doesn’t want to drive, what can we do? All home delivery options are handled by the warehouse; local door-to-door delivery isn’t an option. A possible solution is to move inventory between store locations. Would the customer accept a shorter delay if the item was delivered to the store closest to them? If we can deliver the item to the local store for pickup and reduce the delay enough, there’s a good chance we can save that order and keep the customer happy.
With MuleSoft for Agentforce, we can replicate this scenario across a variety of different third-party applications using topics.
In the graphic, Agentforce has been assigned a topic called “Delayed Order Resolution” and has all the appropriate actions and instructions in third-party systems to save that order.
So, if an order is delayed due to a warehouse issue, the first thing Agentforce does is scan Salesforce to cross-check for local stores and then confirm in their inventory system (Netsuite in this scenario) that the item is in stock.
Armed with this information, Agentforce can offer to hold the item for pickup in the drivable location. If that’s acceptable, all we need to do is place a hold on that item in the inventory system at that location.
If this isn’t an acceptable resolution to the customer, the agent then has the capacity to organize an inter-store inventory movement and offer that as the alternative. If the customer accepts, all that’s left is to generate the inventory movement request and update the original order.
That’s just one simple but powerful example of how MuleSoft for Agentforce can augment an already powerful Agentforce process and uplevel it to encompass the business-critical applications and IT infrastructure to create a better outcome for the customer.
The relationship between admins and MuleSoft developers continues to evolve as part of the transition to the agentic world. Admins will work with their MuleSoft counterparts to define the capabilities Agentforce will require outside of the Salesforce ecosystem. MuleSoft developers will evangelize integration resources to admins so that they can fully comprehend the art of what’s possible–and when it comes to Agentforce, that’s quite a lot.
Want to learn more? One of the best ways to learn about MuleSoft for Agentforce: Topic Center is to see it in action in this spotlight demo video. Attending an Agentforce World Tour near you is a great place to start, as is TDX where MuleSoft for Agentforce will be a key topic.
MuleSoft for Agentforce isn’t just a single tool, but a solution that spans both MuleSoft and Salesforce, designed to support the foundation of actionable AI. Users are already seeing the benefits. Adam Anzuoni, Chief Technology Officer at Fundwell, shared, “To truly empower our agents, it’s essential that Agentforce is grounded in the data from our core business systems, securely integrates and scales with platforms like Salesforce, and connects seamlessly to our broader IT infrastructure.” He added, “By giving agents real-time access to reliable data, we enable data-driven actions at every stage of agentic workflows. With MuleSoft for Agentforce, data from MuleSoft APIs can significantly enhance agent capabilities and elevate customer experiences.”
You can head over to MuleSoft.com to learn about everything MuleSoft for Agentforce.
Chicken, meet egg As admins, we’re often confronted with a classic conundrum: How do you convince an organization to adopt new technology without having the technology to demonstrate it in the first place? This is compounded when the technology you want to demonstrate requires other technologies you may not have access to show. For example, […]
Artificial intelligence (AI) agents are intelligent systems that can reason and act on behalf of customers and employees. But to realize their full potential, we need to test and configure agents without disrupting live production environments. That’s where you as an admin come in. Building an effective AI agent isn’t just about deploying advanced models—it’s […]
Have you ever wished you could give your users exactly the right guidance at just the right moment? That’s the magic of Next Best Action! Whether it’s helping a sales rep spot the perfect opportunity for a follow-up or guiding a service agent with a recommended solution, Next Best Action allows you to create custom […]
