Speed, speed, speed. That’s what almost every AI development tool will lead off with. It makes sense. Speed won the cloud era. Speed is the revolutionizing value prop of AI across almost every sector. But that means speed is now the expectation.
Scalability is the next great frontier of AI, and it’s much harder to achieve.
One of the most consistent pieces of feedback we hear from integration teams is this: AI code generation sounds great in theory, but in practice the output is too generic. New code is always the first resort – ignoring the connectors, templates, and assets your team has spent years building. So while it accelerates delivery initially, it creates a new category of manual work: rewriting generated code in an artificially ballooning codebase to conform to standards the AI never knew existed.
This goes against the ethos of composability that MuleSoft has championed for decades, and we firmly believe it represents the gap between AI speed and enterprise scalability. This is why we’re excited to announce the general availability of three new capabilities within MuleSoft Vibes: Custom Connector, Template, and Exchange Reuse Support.
With these updates, MuleSoft Vibes actually understands your enterprise – making suggestions for reuse and generating code that is production-ready from the start, cementing it as the best place for architects, developers, admins, and ops to build with MuleSoft and AI.
Why generic AI code generation falls short in enterprise environments
Most AI coding tools operate in a closed world. They understand public connectors, standard APIs, and generic project structures. For many use cases, that’s fine. For enterprise integration, it’s a fundamental limitation.
MuleSoft customers operate in some of the most strictly regulated and technically complex environments in the world. That means building with proprietary connectors for internal systems, starter templates that encode years of architectural decisions, and Exchange assets that represent reusable building blocks across the business.
When an AI tool has no awareness of any of that, generated code becomes a starting point at best and a liability at worst – particularly in industries where naming conventions, security policies, and error handling are non-negotiable. The productivity gains promised by AI coding tools quietly evaporate.
What’s generally available now?
Let’s take a detailed look into what’s now generally available with MuleSoft Vibes.
Template support
Template support allows MuleSoft Vibes to generate applications using your organization’s own approved starter templates. Rather than generating from a blank slate using generic patterns, Vibes now builds from the architectural foundations your team has already established and validated. This means every generated app inherits your organization’s standards for structure, security, logging, and error handling before a developer writes a single line of additional code. For architects and platform teams, it’s a way to enforce consistency at scale without slowing delivery.
Exchange reuse support
Exchange Reuse Support gives Vibes awareness of the assets your organization has published to Anypoint Exchange. When generating integrations, Vibes can now identify and incorporate relevant Exchange assets: APIs, connectors, fragments, and other reusable components – directly into the generated output.
This directly aligns with the ethos MuleSoft customers have adopted for the past two decades: Teams stop rebuilding what already exists. Assets that have been invested in and tested get used more consistently. And new integrations are generated with a foundation that reflects how your organization has already chosen to solve common problems.
Custom connector support
Custom connector support extends Vibes’ code generation to include your organization’s proprietary and REST-based connectors. Previously, generated code could only consume publicly available connectors – meaning any integration that touched a custom system had to be wired up manually after generation.
With Custom Connector Support, developers can describe an integration in natural language and have Vibes generate code that incorporates the right custom connectors from the start. The gap between what the AI produces and what can actually go to production gets significantly smaller.
Built for the entire organization
These three capabilities are designed to work together, and their combined effect goes beyond individual developer productivity. When Vibes generates code that respects templates, reuses Exchange assets, and knows your custom connectors, the output is inherently more consistent across teams. New developers onboard faster because the AI guides them toward organizational standards rather than away from them. Architects get governance without friction. Platform teams see their investments in reusable assets start to compound.
This is what it means for AI code generation to scale in an enterprise context. Faster individual output, sure, but better organizational outcomes for the entire IT team.
Getting started
Custom Connector Support, Template Support, and Exchange Reuse Support are now available within MuleSoft Vibes for all Anypoint customers with a connected, Einstein-enabled Salesforce org. To learn more, stay up to date with everything MuleSoft and AI by subscribing to our newsletter, Technically Speaking.
