The shift took years. Salesforce’s purchase of MuleSoft in 2018 moved integration closer to the center of the CRM platform, later releases made APIs easier to expose inside Salesforce, and agent use has raised the cost of weak API discovery and control. For companies reviewing their integration architecture now, the useful question is how to reduce repeated connection work while keeping APIs understandable and governable as demand rises.
The 2018 acquisition put integration inside Salesforce’s platform strategy
Salesforce completed its MuleSoft acquisition on May 2, 2018 after agreeing to an enterprise value of about $6.5 billion. At the time of the March agreement, MuleSoft had more than 1,200 customers, and Salesforce said Anypoint Platform would become part of Salesforce Integration Cloud. The move tied Salesforce’s CRM expansion to a broader application-network model that could connect cloud services with systems running on premises. Salesforce’s 2018 MuleSoft acquisition announcement
That timing matters because many enterprises were still carrying large sets of direct system connections. A new CRM process could require another custom interface to an ERP, billing platform, warehouse, or legacy database, and each new dependency added testing and maintenance work. The appeal of Data Integration Solutions in that setting is the chance to build integration around reusable APIs and shared connection patterns instead of treating every request as a separate wiring job. The client page places API-led design within an integration process that also covers mapping, testing, deployment, and architecture.
API reuse changed what teams expected from integration work
The years after the acquisition pushed integration teams toward reusable API assets. Once a system capability has a stable interface, another team can call that interface rather than rebuild the same connection from scratch. That reduces duplicated engineering work, but it also creates a new responsibility: the API needs clear ownership and access rules, along with documentation and a change process that other teams can trust.
This is where MuleSoft Integration Services become relevant to delivery speed. The service decision involves identifying capabilities that deserve reusable interfaces and deciding which older connections can remain in place. API management then has to make those assets easier to find and maintain. Later projects can begin with working interfaces rather than another round of custom connection design.
The 2024 release plan pulled APIs closer to Salesforce workflows
A further change became visible at Dreamforce 2024. Salesforce announced MuleSoft features that would bring APIs directly into Salesforce, including API Catalog and low-code integration, with general availability planned for the first quarter of 2025. API Catalog was designed to surface APIs from across the Salesforce ecosystem, including MuleSoft and Heroku, so they could be used in flows and Agentforce actions. Salesforce’s 2024 MuleSoft announcement
That change affected the audience for integration assets. APIs were useful to a wider group than integration developers working inside a separate platform console. Salesforce administrators and teams building automation could call governed services from the tools where they already worked, which increased the value of having a well-organized API set. For companies using Salesforce across several functions, Salesforce Data Integration Solutions therefore need to account for reuse inside Salesforce as well as the original task of moving records between systems.
The 2025 API data showed why bottlenecks moved beyond coding
By 2025, API-first development had become common enough that the pressure was shifting from basic adoption to operational discipline. Postman’s 2025 State of the API survey covered more than 5,700 people across development and architecture roles, including executives, and found that 82% of organizations had adopted some level of an API-first approach. It also found that only 24% of developers were actively designing APIs with AI agents in mind, while 93% of teams reported some form of API collaboration difficulty. Postman’s 2025 State of the API Report
Those numbers explain why more APIs don’t automatically mean faster delivery. A large catalog with weak documentation or unclear ownership can slow a project because engineers still have to discover what exists and confirm which version to use. Access approval can add another wait when responsibility for an API isn’t obvious. The 24% figure also shows a design gap as agents become API consumers, because an interface built mainly around human assumptions may lack the predictable contracts that automated callers need.
The practical response is to treat Salesforce Data Integration as an architecture and governance question as much as a data-movement task. Teams need to decide which business capabilities deserve reusable APIs and how those APIs will be found. They also need change rules that protect downstream consumers when an interface evolves. Settling those questions before implementation removes delays caused by duplicate interfaces and unclear dependencies.
The 2026 releases made agent access part of API management
MuleSoft’s 2026 product changes make the next stage visible. In May, API Experience Hub added support for publishing agentic assets such as MCP servers and agents beside other service types, and portal capacity moved to 7,000 assets by default. That is a sharp increase from May 2024, when the release notes listed a maximum of 350 APIs, and from February 2025, when the limit rose to 1,000 for supported portals. MuleSoft API Experience Hub release notes
The numbers show a product moving toward much larger service catalogs, while the new asset types extend those catalogs beyond conventional APIs. This raises the cost of weak naming and ownership because more automated consumers can depend on the same services. Lifecycle rules and access control also carry more weight when a change can affect application workflows as well as agent actions. The delay may now come from finding and approving the right capability rather than writing the endpoint itself.
The next integration decision should focus on reuse before volume
The timeline points to a clear decision for Salesforce teams. Integration work moved from building individual connections toward managing reusable APIs, and the arrival of agents increases the number of consumers that may depend on those APIs. The safer way to move faster is to reduce duplicate connection work while keeping service ownership and access rules clear enough for application workflows and agent actions.
An integration review should therefore start with e
