MuleSoft is Salesforce's integration and API management platform. It connects applications, data and devices, whether they run in the cloud or on-premises, through reusable APIs. Its core product, Anypoint Platform, gives teams one place to design, build, deploy, secure and monitor those APIs and integrations, including the ones that link Salesforce to ERP and legacy systems.
This guide explains what MuleSoft does, what each part of Anypoint Platform is for, how API-led connectivity works, how MuleSoft fits with Salesforce and Agentforce, what it costs, and when a lighter tool is the better choice. Every product fact below was read on MuleSoft, Salesforce or Trailhead pages on 11 November 2026. If you are planning a program, see our MuleSoft services page.
What is MuleSoft, in plain terms?
MuleSoft's own definition is "a Salesforce-owned platform that allows businesses to securely connect applications, data, and devices." The company was founded in 2006. Salesforce announced its agreement to acquire MuleSoft on 20 March 2018 at an enterprise value of about $6.5 billion.
In practice, MuleSoft is the layer that sits between your systems. Instead of Salesforce calling SAP directly, and the website calling SAP separately, and the data warehouse pulling from both, each system is exposed once through a managed API. Other applications then reuse that API. MuleSoft calls the result an application network.
The name you will see most is Anypoint Platform, the product that holds the design tools, the runtime, the hosting options and the API management. The engine that actually runs integrations is the Mule runtime engine, which the MuleSoft documentation describes as "a lightweight integration engine" that runs Mule applications. Version 4.13 was marked as the latest release in the docs when we checked.
What is MuleSoft used for?
MuleSoft groups its uses into system integration, AI and API management, workflow automation and agentic AI. The table shows what these, plus two related patterns, look like on a typical enterprise Salesforce program.
| Use case | What MuleSoft does | Salesforce example |
|---|---|---|
| System integration | Moves and transforms data between SaaS, ERP, databases and legacy systems | Orders, invoices and credit status from SAP or Oracle shown on the Salesforce account |
| API management | Publishes, secures and governs APIs with policies, rate limits and analytics | One governed customer API used by Sales Cloud, a portal and a mobile app |
| Workflow automation | Triggers multi-step processes across systems when an event happens | A closed-won opportunity creates the customer and contract in the ERP |
| Data unification | Combines records from several sources into one consistent view | Policy, claims and billing data assembled for a Financial Services Cloud advisor |
| Document processing | Extracts structured data from PDFs and images with Intelligent Document Processing (IDP) | Supplier invoices or application forms turned into Salesforce records |
| Agentic AI | Gives AI agents governed access to enterprise systems through MCP and A2A | An Agentforce agent checks order status in the ERP before answering a customer |
What are the main parts of MuleSoft Anypoint Platform?
Anypoint Platform is a set of tools that cover the full life of an API. The descriptions below come from MuleSoft's enterprise integration page and the MuleSoft documentation.
| Component | What it is for |
|---|---|
| Anypoint Code Builder | The IDE developers use to design, build and deploy APIs, integrations and automations |
| Anypoint API Designer (Design Center) | Tooling for writing and testing API specifications before implementation |
| Anypoint Exchange | "The marketplace to discover and share APIs, connectors, examples, and templates" |
| DataWeave | MuleSoft's language for data transformation, used to map formats such as JSON, XML, CSV and flat files |
| Mule runtime engine | Executes and routes integration flows and data transformations |
| CloudHub | MuleSoft's fully managed, multi-tenant cloud platform for running Mule applications |
| Runtime Fabric | Container service for running Mule applications on Docker and Kubernetes in AWS, Azure, GCP or your own data center |
| API Manager | Secures, manages and reports on APIs from one control plane |
| Omni Gateway | API gateway for any environment; MuleSoft calls it "the next evolution of Flex Gateway" |
| Anypoint Monitoring | Dashboards, alerts and log access for API and integration performance |
Connectors are the other building block. MuleSoft's enterprise integration page says the platform is underpinned by "over 1,500 pre-built connectors and integrations," while other MuleSoft pages say "hundreds." Either way, common systems such as Salesforce, SAP, Workday, ServiceNow and major databases have maintained connectors, so teams do not write raw API clients for them.
How does MuleSoft work?
A MuleSoft integration follows the same path from idea to production. This is the sequence we see on most programs:
- Design the API contract. The team defines the API in RAML or OpenAPI: its resources, fields and errors. The contract is published to Exchange so consumers can review it before any code exists.
- Build the Mule application. In Anypoint Code Builder, developers assemble flows from connectors (for example, Salesforce and SAP) and processors that route, filter and handle errors.
- Transform the data. DataWeave maps one system's format to another, such as SAP order lines into the JSON shape Salesforce expects.
- Deploy the runtime. The application runs on CloudHub, on Runtime Fabric, or on a self-managed Mule runtime, depending on where the target systems sit.
- Apply policies. API Manager and the gateway add authentication, rate limits and access rules without changing the application code.
- Monitor and reuse. Anypoint Monitoring tracks performance and errors, and the API stays in Exchange for the next project to reuse.
Step six is where most of the value comes from. Salesforce's Anypoint Platform page claims teams can deliver projects "five times faster with reuse versus custom code." That is a vendor figure, but the logic holds: the second project that needs customer data calls an existing API instead of building another connection.
What is API-led connectivity?
API-led connectivity is the architecture MuleSoft recommends. Trailhead's API-led connectivity unit says it "organizes APIs into three distinct layers, each serving a specific purpose."
| Layer | Trailhead's description | Example on a Salesforce program |
|---|---|---|
| System APIs | "Connect directly to core systems like databases, ERPs, or CRMs" | An SAP Orders System API and a Salesforce Accounts System API |
| Process APIs | "Orchestrate and process data from multiple System APIs and other Process APIs" | An Order Status Process API that joins SAP orders to the Salesforce account and adds shipping data |
| Experience APIs | "Deliver data in formats optimized for user interfaces or devices" | One Experience API for the Experience Cloud portal and another shaped for an Agentforce action |
The point of the layers is change isolation. If the company replaces its ERP, only the System API changes. The Process API, the portal and the agent keep working. Trailhead puts the benefit simply: "Teams can repurpose existing APIs, reducing duplication and accelerating development."
Not every integration needs all three layers. A one-way nightly load into a data warehouse may need a single API. We decide the layering per domain, and keep three layers for data many consumers depend on, such as customer, product, order and payment.
How does MuleSoft differ from point-to-point integration?
MuleSoft's own page says "traditional integration relies on rigid, point-to-point connections." Point-to-point is fine for two systems. It becomes expensive at ten, because each new system adds connections to every system it touches.
| Factor | Point-to-point (custom code, Apex callouts) | MuleSoft API-led |
|---|---|---|
| Reuse | Each connection is built for one consumer | APIs in Exchange are reused by new consumers |
| Impact of replacing a system | Every connection to it must be rewritten | Only the System API changes |
| Security and policies | Set separately in each integration | Applied centrally through API Manager and the gateway |
| Monitoring | Scattered logs across systems | One monitoring view for APIs and flows |
| On-premise systems | Need custom network exposure or VPN work per integration | Runtime Fabric or standalone runtimes run next to the system |
| Upfront cost | Low for the first integration | Higher: licensing, platform setup and design standards |
| Best fit | A few simple, stable connections | Many systems, many consumers, regulated or hybrid estates |
For a wider view of the options, including when Salesforce-native tools are enough, read our guide to Salesforce integration patterns.
How do MuleSoft and Salesforce work together?
There are three main ways MuleSoft meets Salesforce today.
- The Salesforce Connector. MuleSoft's Salesforce Connector (version 12.1 was the latest when we checked) works with the Apex SOAP API, Apex REST API, Bulk API, Metadata API and Streaming API. It lets Mule applications react to Salesforce events and read or write records at volume. The docs note it does not provide access to the Chatter or Tooling APIs.
- MuleSoft for Flow: Integration. A low-code option for admins. MuleSoft says it brings 200+ native, low-code connectors into Salesforce Flow Builder, so simple integrations can be built without a developer.
- Agentforce. MuleSoft supports Model Context Protocol (MCP) and Agent-to-Agent (A2A) to give AI agents the tools they need in ERPs, CRMs and databases. MuleSoft states that "MuleSoft Agent Fabric and Agentforce Actions embed these capabilities directly into Agentforce." Agent Fabric is MuleSoft's control plane for discovering, orchestrating and governing agents across platforms.
The Agentforce point is the one most top-ranking explainers miss. An agent that can only see Salesforce records can answer questions. An agent that can call a governed Order Status API can resolve the case. The quality of your APIs sets the ceiling for what your agents can do, which is why we review the integration layer early in our Agentforce implementation programs.
Where can MuleSoft applications run?
MuleSoft is unusual among integration platforms in how many hosting choices it gives you. The Mule runtime documentation lists these options:
| Option | Who manages the infrastructure | When it fits |
|---|---|---|
| CloudHub 2.0 and CloudHub 1.0 | MuleSoft | Cloud-to-cloud integrations and teams that do not want to run servers |
| Runtime Fabric | You, on containers in your data center or a third-party cloud | Data residency rules, on-premise systems, Kubernetes standards |
| Hybrid standalone runtimes | You, connected to Anypoint's cloud management | Existing servers that should still be managed centrally |
| Standalone runtimes | You, with no required connection to the Anypoint control plane | Isolated or highly restricted networks |
For US banks, insurers and healthcare organizations, this flexibility is often the deciding factor. Data can stay inside the network while the APIs are still governed from one place.
How much does MuleSoft cost?
MuleSoft's pricing page (read 11 November 2026) gives these facts:
- Packages start at $2,000 per month, billed annually.
- Usage is measured in Mule Credits, which MuleSoft defines as a shared unit of consumption across eligible MuleSoft capabilities. Per-credit rates fall as usage grows, by up to 90%.
- Pre-Purchase: buy a package and a block of credits up front, with no minimum credit purchase.
- Pre-Commit: commit to a yearly baseline and draw it down monthly, with a $50,000 minimum annual commitment.
- Unlimited: a custom fixed price with uncapped credits.
- A 30-day free trial gives access to the full portfolio.
License cost is only part of the total. Plan for platform setup, API design standards, CI/CD pipelines and people who know DataWeave. The cost of a MuleSoft program is usually driven by the number and complexity of the systems involved, not by the license alone.
When is MuleSoft the right choice, and when is it not?
MuleSoft is a strong fit when several of these are true:
- Salesforce has to exchange data with three or more core systems, such as ERP, billing, data warehouse and a policy or core banking platform.
- The same data (customer, order, product) is needed by several consumers: Salesforce, a portal, a mobile app, analytics and AI agents.
- Some systems sit on-premises or under data residency rules.
- Security teams want central API policies, audit and monitoring.
- Agentforce agents need to take actions in systems outside Salesforce.
A lighter option is usually better when you have one or two simple connections, low volumes and no reuse. Salesforce-native tools such as Flow with External Services, platform events or an AppExchange connector may be enough. If business teams own most automations, a low-code platform can also fit; we compare one in Workato vs MuleSoft.
How do you get started with MuleSoft?
A first MuleSoft project should prove reuse, not just move data. This is the approach our architects use:
- Map the systems and data owners. List every system Salesforce must talk to, the data it owns, its volumes and whether it is cloud or on-premises.
- Pick one high-value domain. Customer or order data is often the best start, because many teams and channels need it.
- Agree the API standards. Naming, versioning, error formats and security policies, written down before the first API is built.
- Choose the hosting model. CloudHub, Runtime Fabric or hybrid, based on where the systems sit and your compliance needs.
- Build System, Process and Experience APIs for that domain. Publish them to Exchange with documentation.
- Set up CI/CD and monitoring. Automated deployment, alerts and dashboards from day one.
- Reuse on the second project. Measure how much of the second integration came from existing APIs. That number is the business case for scaling.
Techila Global Services is a Salesforce Summit Partner headquartered in Frisco, Texas, with 350+ Salesforce professionals, 500+ Salesforce certifications and 700+ projects delivered since 2012. Our MuleSoft services and Salesforce integration services teams run architect-led programs from first design through managed support.
Frequently asked questions
Is MuleSoft part of Salesforce?
Yes. MuleSoft was founded in 2006 and Salesforce announced its agreement to acquire the company on 20 March 2018, at an enterprise value of about $6.5 billion. MuleSoft now describes itself as a Salesforce-owned platform, and it is sold by Salesforce alongside the Salesforce clouds.
Is MuleSoft an ESB, an iPaaS or an API management tool?
It covers all three jobs. The Mule runtime engine does the routing and transformation work an enterprise service bus does, Anypoint Platform adds managed cloud hosting and design tools, and API Manager plus Omni Gateway handle API security, policies and governance from one control plane.
Do you need to code to use MuleSoft?
For Anypoint Platform, mostly yes: developers build Mule applications in Anypoint Code Builder and write transformations in DataWeave, MuleSoft's data transformation language. Admins and business teams can use MuleSoft for Flow: Integration, a low-code tool that puts pre-built connectors inside Salesforce Flow Builder.
How much does MuleSoft cost?
MuleSoft's pricing page lists packages starting at $2,000 per month, billed annually, with usage measured in Mule Credits. Buying models are Pre-Purchase, Pre-Commit with a $50,000 minimum annual commitment, and Unlimited at a custom fixed price. A 30-day free trial is available.
Can MuleSoft connect Salesforce to SAP, Oracle or on-premise systems?
Yes. MuleSoft provides pre-built connectors for common ERP, database and SaaS systems, and Mule applications can run on-premises or in your own cloud through Runtime Fabric or standalone runtimes. That lets Salesforce reach systems that are not exposed to the internet.
What does MuleSoft add to Agentforce?
Agents need to read and update data that often lives outside Salesforce. MuleSoft supports Model Context Protocol (MCP) and Agent-to-Agent (A2A), and MuleSoft says Agent Fabric and Agentforce Actions bring these capabilities into Agentforce, so agents can act in ERPs, databases and legacy systems under governed APIs.
Talk to a MuleSoft architect
Tell us which systems Salesforce needs to reach and what your teams or agents should be able to do with that data. A MuleSoft architect will review your landscape and suggest an API-led design and a hosting model that fits.