MuleSoft Connect AI 2025: The journey from digital... to agentic

From digital… to agentic

Looking at the role of AI agents being deployed in current enterprise IT stacks, MuleSoft preluded the event by showcasing a whitepaper entitled From Digital To Agentic Transformation and saying that (just as human employees rely on knowledge repositories and guidelines), AI agents require a flexible and adaptable IT infrastructure to thrive.

“Organisations that have adopted composable architectures, which emphasise modularity, flexibility and interoperability, are not only able to deliver projects faster but are also uniquely positioned to leverage the power of AI,” noted the company. “For over two decades, MuleSoft has been at the forefront of helping organisations implement these composable architectures to drive digital transformation. In the agentic era, the key to success lies in simplifying complexity, standardising access and automating business processes. MuleSoft provides the essential tools and support to build this robust foundation, enabling businesses to fully capitalise on the AI revolution.”

As an overview of some of the key MuleSoft tools, Salesforce API Catalog provides a single repository of APIs from Salesforce, MuleSoft and Heroku embedded directly in Salesforce. MuleSoft Direct for Data Cloud enables organisations to bring essential information from external systems into Data Cloud, with the aim of fueling AI applications across Salesforce. By using pre-built applications to ingest unstructured data, organisations can augment AI prompts, resulting in more accurate and reliable outputs from large language models (LLMs).

MuleSoft AI Chain is, an open source project that offers interaction with LLMs, vector databases and APIs, enabling software application developers to build their own agent or copilot. 

According to the company, “The MuleSoft AI Chain project aims to address [agent building] challenges for MuleSoft users, while simultaneously making it easy to extend copilots and agents with existing MuleSoft APIs. It offers a suite of connectors designed for MuleSoft developers, enabling seamless interaction with LLMs, vector databases, APIs and token management. With these connectors, customers can implement a wide range of AI use cases, including RAG for both online and offline scenarios, agentic workflows and knowledge bases.”

Agent Actionability

At the MuleSoft Connect AI event itself, the company hosted key sessions on agentic AI usability, connectivity and composability – it’s what MuleSoft calls agent actionability today.

As well as sessions devoted to MuleSoft for Agentforce. The company says that developer teams can now upskill the digital workforce with MuleSoft and Agentforce. They can do this by enabling AI agents to proactively take action in any system or workflow, all with the proper guardrails in place. MuleSoft for AI supports the transition from composability to actionability. 

MuleSoft also used its event to showcase how low-code tools can connect siloed systems, orchestrate end-to-end processes and accelerate automation across a business. With intelligent agents handling repetitive tasks and triggering actions, the promise is that firms will be able to save time and boost efficiency (creating a truly connected enterprise) all without writing a single line of code.

News from the company this month saw it come forward with Model Context Protocol (MCP) support, to transform applications or APIs into tools and resources for AI agents. MuleSoft Agent2Agent (A2A) support enables agents to communicate, collaborate and take coordinated actions across the enterprise with governance. Also new are generative AI tools in MuleSoft’s Integrated Developer Environment (IDE) allow developers to build integrations faster and accelerate project development.

The company also underlined the fact that the recent launch of Salesforce Agentforce 3, which includes native MCP support, data and software engineering teams can also use MuleSoft’s new features to turn APIs into MCP servers and extend Agentforce with new opportunities for agent interoperability without compromising control.

“MuleSoft’s agent orchestration marks the next evolution in enterprise integration. It’s how organizations will power truly intelligent, actionable interactions by unifying APIs, agent actions, governance, and multi-agent coordination,” said  Andrew Comstock, SVP and GM of MuleSoft, Salesforce

MCP is rapidly gaining momentum as a standardised, open-source framework for enabling secure bidirectional communication between LLMs or AI agents and external systems like Salesforce. MuleSoft MCP Support enables integrations with LLMs and enterprise systems and provides agents with the ability to take action in any system with governance.

“MuleSoft has enabled us to build a composable architecture and data foundation needed to scale AI across AstraZeneca – from HR to R&D to commercial. Its commitment to evolving industry protocols and the crucial semantic layer ensures secure and scalable orchestration of agent-to-agent and agent-to-system actions, making our strategic partnership with Salesforce and MuleSoft vital for our ambitious AI journey,” said Wayne Filin-Matthews, chief distinguished architect, AstraZeneca.

Product availability

Due to the large number of products and services launched right now, it’s worth covering off availability. MuleSoft MCP Connector is expected to hit general availability (GA) in July 2025, as is Mulesoft A2A Connector, MuleSoft Flex Gateway: MCP Support is generally available as its Anypoint Code Builder: AI IDE Support.  Einstein for Anypoint Code Builder: API Spec is expected to GA imminently, as its Einstein for Anypoint Code Builder: Dataweave. Anypoint Exchange: AI-Generated Documentation is generally available.