How MuleSoft’s AI-Powered Composability Will Help Organizations Prepare for an AI Future

Most business leaders expect that every enterprise application will include predictive or generative AI capabilities in the future. However, to truly leverage AI, businesses need to deeply integrate it into their current applications and workflows. The benefits of AI — generating intelligent insights, content, and automation through AI agents — need to happen in the natural flow of work, not as a separate, siloed process. 

MuleSoft’s suite of integration, automation, and API management solutions helps companies build a strategy to fully leverage AI in their organization now and in the future. MuleSoft’s AI-powered composability solution combines an AI-centric methodology with tools that help developers and business teams integrate AI into their applications and workflows via building blocks. 

With AI-powered composability approaches from MuleSoft, companies can:

• Build a future-ready foundation to connect any system or LLM and create AI-powered customer experiences.
• Gain efficiencies with intelligent tooling so developers and business teams can accelerate productivity with AI-powered integration and automation solutions. 
• Secure AI-powered experiences by giving companies governance and access control for trusted AI architectures and applications. 
• Power AI agent actions to ensure data-driven responses and take action in external systems using natural language prompts.

Here, Param Kahlon, Salesforce EVP and GM of Automation and Integration, goes deeper into how AI-powered composability is crucial for organizations aiming to optimize their use of AI now and in the future.

Q. How can AI-powered composability support companies with their AI transformation?

Composability, or the way individual building blocks can be used to build more intricate systems and applications in a technology stack,​​ empowers organizations to rapidly adopt new standards and technologies because it is inherently flexible. MuleSoft’s platform is designed for customers to build composable architectures — helping them scale and flex as technologies and standards evolve using a diverse set of integration, automation, and API management capabilities.

AI-powered composability is MuleSoft’s solution to help organizations build a foundation for the new AI era. By combining an AI-centric methodology with tools that help developers and business teams create AI-powered applications via building blocks, organizations can combine and customize the assets they need to easily meet any AI requirement for their current and future applications, building a secure and adaptable foundation. 

AI-powered composability is MuleSoft’s solution to help organizations build a foundation for the new AI era.

Param Kahlon, Salesforce EVP and GM of Automation and Integration

Q. How does MuleSoft support building AI-powered experiences?

MuleSoft’s diverse product portfolio includes Anypoint Platform, for integration and API management, Robotic Process Automation (RPA), Intelligent Document Processing (IDP), Composer, and Salesforce Flow. Together, these products provide the foundation for organizations to create AI-enabled apps, manage and secure LLM APIs, and allow AI agents to act on third-party data — all critical steps in building an AI enterprise. And with Salesforce Einstein, we have built AI into our tools for developers and business teams to integrate and automate faster. 

Q. How can customers tap into this AI-powered composability concept?

We have customers across all industries, but let’s look at an automotive manufacturing company that is now selling cars directly to consumers. 

As they navigate this massive shift in their business model, this company sees AI as an opportunity to help differentiate the experiences they’re providing to their new customers. They want an AI-powered experience in driver dashboards, in their mobile apps, and in their web applications. They also want to make sure that their cars are intelligent. 

To collect telemetry data, such as driving behaviors or the state of the engine, the manufacturer can integrate their systems using an API-led methodology. Once the foundation is set, AI-powered composability comes in. They can continue to build reusable building blocks such as APIs, integrations, or AI-powered automation flows and integrate LLMs into their architecture. This, in turn, will enable them to provide intelligent actions such as proactive maintenance alerts to their customers.

In addition, MuleSoft enables organizations to harness the power of AI to automate processes and connect systems faster to help increase worker productivity and help meet increasing customer expectations.

Q. Security is top of mind for organizations. What should they think about when integrating LLMs into their applications? 

Customers are concerned about who gets to see their data and how data access will be monitored and controlled across the organization. And they’re really concerned about data leaving the corporate boundary. Is that data protected? Encrypted? Masked? 

MuleSoft believes LLM APIs need security and governance just like any other API. We have customers using our API gateways (Anypoint Flex Gateway and Mule Gateway) as LLM gateways with custom-made policies to help them secure and manage their APIs. The gateways help manage the data that the LLM APIs are calling, such as encrypting and masking personally identifiable information (PII).

For example, with an API gateway, financial institutions can ensure that data stays within their trusted boundaries by developing a custom policy that checks for sensitive customer information before data is shared with a third-party LLM.

MuleSoft also has built-in governance rulesets that provide automated conformance checks to help support industry-specific and corporate compliance. For example, banks can include a standard procedure for managing LLM APIs to ensure proper user authentication before gaining access to generative AI capabilities.

We also provide monitoring capabilities for the entire API and integration lifecycle, from LLM API governance to AI request tracing. For example, the new Anypoint API Insights allows developers to take action on LMM APIs that don’t meet company conformance requirements, creating a trusted AI foundation on which to build applications. 

Q. How is MuleSoft leveraging AI in its own tools?

We want to make sure that everybody in the organization can speak and communicate in natural language to generate the code to generate integrations, to generate flows — all with prompts. 

As a Salesforce company, MuleSoft has embedded Einstein into our products. Einstein for Anypoint Code Builder (in beta and generally available in August), uses natural language prompts so developers can quickly build integrations, while Einstein for Flow (in pilot and in beta in July) enables Salesforce admins to automate business processes. And, with the recent launch of Intelligent Document Processing (IDP), users can tap Einstein for IDP (in pilot) to extract data from documents such as a driver’s license or invoice. These prompts go through the Einstein Trust Layer to ensure that your data reaches the LLM.

Solutions like these help anyone increase productivity and focus on more value-oriented work. 

Q. How can companies maximize the potential of today’s AI assistants?

To be meaningful, AI assistants need to go beyond just executing a search or answering questions. They need to take action with conversational prompts such as updating an employee record or vacation request, creating an order, or updating a customer’s bill. If implemented correctly, AI should take on the repetitive, mundane tasks that humans do currently.

To be meaningful, AI assistants need to go beyond just executing a search or answering questions. They need to take action with conversational prompts such as updating an employee record or vacation request, creating an order, or updating a customer’s bill.

Param Kahlon, Salesforce EVP and GM of Automation and Integration

MuleSoft technology is helping bring these capabilities to life in Einstein Copilot, Salesforce’s predictive and generative AI assistant. Using APIs and integrations built with MuleSoft Anypoint Platform (coming soon) and Salesforce Flow, Copilot can extend its capabilities to leverage data from third-party systems. This enables Copilot to create actionable workflows and data-rich responses. 

For example, AI agents can create an order in a back-office system to give a credit to the customer or alert a supplier to procure inventory ahead of an urgent order that’s being delivered. This is possible to do in a conversational language because the APIs are available to automate end-to-end business processes. 

As companies look to become an AI enterprise — a company that has transformed its operations by adopting AI technologies that both support and autonomously perform tasks as an extension of their teams — AI-powered composability will be critical to drive AI success now and in the future. 

More information:

• Read more news from MuleSoft Connect:AI 
• Learn more about MuleSoft AI solutions

Any unreleased services or features referenced here are not currently available and may not be delivered on time or at all. Customers should make their purchase decisions based upon features that are currently available.