AI INTEGRATIONS
If everything you’ve read about Sparx EA AI integration has been about Microsoft, this page is for you.
The integration narrative in the EA space has been Microsoft-heavy: Copilot, Power BI, Teams, Fabric. For organizations standardized on Salesforce’s ecosystem, that narrative doesn’t apply. The Microsoft path isn’t your path. But the underlying integration capability is parallel, because EA GraphLink doesn’t care which ecosystem you’re on.
One EA GraphLink deployment. Three Salesforce integration points:
You don’t need all three. Most organizations start with one or two and add the third only if the use case is confirmed. But the path exists in full, and one EA GraphLink deployment supports all three simultaneously.
The Microsoft and Salesforce integration paths are architecturally equivalent. Same EA GraphLink foundation. Same technical interfaces (GraphQL for BI dashboards, MCP for AI and integration tools). Different platforms, different ecosystems.
| Use Case | Microsoft Path | Salesforce Path |
|---|---|---|
| BI Dashboards | Power BI (GraphQL) | Tableau (GraphQL) |
| AI Assistant | Microsoft Copilot (MCP) | Salesforce Agentforce (MCP) |
| Data Integration | Microsoft Fabric (MCP) | MuleSoft Fabric (MCP) |
The interface is the same. The target platform is different. For Salesforce-centric organizations, this means the EA integration story applies to you: it’s been described in Microsoft terms, but the capability is identical.
Everything starts here. EA GraphLink is the connectivity layer that transforms the physical Sparx EA repository schema: via your MDG Technology definition: into a queryable interface accessible to external tools. Without EA GraphLink, no external integration (Salesforce or otherwise) is possible.
EA GraphLink exposes two interfaces:
EA GraphLink Foundation is deployed as part of every Connect engagement. It’s the prerequisite for every integration on both the Salesforce and Microsoft paths.
One critical note before any integration work begins: the quality of your MDG Technology definition directly determines the quality of everything downstream. EA GraphLink transforms the physical schema using your MDG definition. If your MDG is poorly governed: stereotypes inconsistently applied, tagged values incomplete, relationships weak: the transformation produces a noisy, unreliable knowledge graph. Tableau dashboards will show incomplete data. Agentforce will return inconsistent answers. MuleSoft will receive poor-quality data.
MDG readiness is assessed in the Discover engagement before Connect begins. If your MDG governance is uncertain, Discover is the right starting point.
Tableau connects to EA GraphLink’s GraphQL interface. The connection is analogous to connecting Tableau to any GraphQL data source: EA GraphLink acts as the data provider, Tableau is the visualization and dashboard platform.
Once connected, Tableau can display live EA repository data in dashboards that auto-refresh when the repository changes. No manual export. No periodic CSV production. No stale snapshots. Architecture data in Tableau stays current because it’s always sourced directly from the repository.
What this provides: BI-quality visualization of architecture data: application portfolios, technology landscapes, capability maps, compliance coverage, lifecycle timelines: in the Tableau environment your business users already know. IT executives, portfolio managers, and governance teams get architecture visibility through the BI tool they’re already using.
The practical setup: Sparx Services configures the EA GraphLink GraphQL endpoint, establishes the Tableau connector, designs the initial dashboard set based on your governance priorities, and validates data accuracy. Ongoing dashboard development happens within your team using standard Tableau skills.
Salesforce Agentforce connects to EA GraphLink’s MCP Server. Agentforce is Salesforce’s AI agent platform: the Salesforce ecosystem equivalent of Microsoft Copilot. With the MCP connection in place, Salesforce users can query the EA repository through Agentforce in natural language, within Salesforce.
A business stakeholder working in Salesforce on a major customer account asks: “What applications support our customer management platform, and what’s their current lifecycle status?” Agentforce queries the EA repository via MCP and returns current data. The stakeholder has architecture context without leaving Salesforce, without an EA login, without requesting a briefing from an architect.
This changes the information flow the same way Copilot integration does for M365 users: but for your Salesforce-using stakeholders. Business development teams, account executives, and program managers who live in Salesforce can now access architecture data where they work.
Governance note: Agentforce is a cloud service. Architecture data queried through Agentforce flows through Salesforce’s infrastructure, not within your local environment. For regulated industries, confirm that this data flow meets your compliance requirements before deploying.
MuleSoft Fabric connects to EA GraphLink’s MCP Server. For organizations where MuleSoft is already the enterprise integration standard, EA repository data becomes a connected data source within the MuleSoft platform.
This enables: automated synchronization of architecture data with other enterprise systems, event-driven integration (EA repository changes trigger downstream workflows), and EA data as an authoritative source in data integration pipelines. Practical examples include synchronizing EA application data to CMDB, connecting EA capability models to HR or finance systems, or including EA architecture data in automated reporting pipelines.
MuleSoft integration is a higher-complexity undertaking than Tableau or Agentforce. It’s worth pursuing only if MuleSoft is already the organization’s integration standard and there’s a confirmed use case for EA data flowing into that integration estate. Don’t add MuleSoft to the scope simply because the option exists.
Start with Tableau if:
BI dashboards are the primary unmet need. Executives and portfolio managers want architecture data in dashboards. The organization already uses Tableau. This is the lowest-friction Salesforce ecosystem integration: straightforward to configure, immediately visible value, no AI adoption required.
Add Agentforce if:
Stakeholder self-service is needed and your organization’s key users are in Salesforce. Natural language access to architecture data inside Salesforce would remove a significant information bottleneck. Business stakeholders are asking architecture questions that currently require architect involvement. Note: Agentforce licensing adds cost: validate the user count and use case before committing.
Add MuleSoft only if:
MuleSoft is already the enterprise integration platform and there’s a specific, confirmed use case for EA data flowing into MuleSoft pipelines. MuleSoft integration is complex and expensive to configure. It creates significant value for organizations with mature MuleSoft estates and clear EA data requirements. For organizations without MuleSoft or without a specific integration use case, this step isn’t justified.
One note on sequencing: Each step above can be done independently. You don’t need to commit to all three at project start. A common pattern is to begin Connect with EA GraphLink Foundation + Tableau, confirm the value, and then add Agentforce in a subsequent phase. This limits initial scope risk while establishing the foundation that makes subsequent integrations faster.
The MDG Technology definition governs what EA GraphLink can do with your repository. This is true for every integration: Microsoft or Salesforce: and it cannot be bypassed.
EA GraphLink reads the physical schema of your Sparx EA repository and transforms it using your MDG Technology definition. The MDG defines: what element types exist, what stereotypes are applied, what tagged values capture, and how relationships are structured. EA GraphLink uses this definition to build the queryable knowledge graph that Tableau, Agentforce, and MuleSoft connect to.
If MDG governance is strong: stereotypes applied consistently, tagged values populated, relationships accurately maintained: EA GraphLink produces a high-quality knowledge graph. Tableau dashboards show accurate, complete data. Agentforce returns reliable answers. MuleSoft receives authoritative data.
If MDG governance is weak: stereotypes inconsistent, tagged values missing, relationships sparse: EA GraphLink produces an unreliable knowledge graph. Tableau shows incomplete or conflicting data. Agentforce returns inconsistent or confusing answers. MuleSoft receives poor-quality data that propagates errors downstream.
MDG readiness is assessed during Discover. Organizations with uncertain MDG governance should complete Discover before committing to Connect.
What Sparx Services provides: BOM (Bill of Materials): a complete list of what software licenses are required, from which vendors, at what scale. We do not resell software.
What you buy directly:
| Software | Where to Buy |
|---|---|
| Sparx EA (Enterprise Architect) | Sparx Systems |
| EA GraphLink | Sparx Systems |
| Kernaro AI Hub (if included) | Sparx Systems |
| Tableau Creator / Explorer | Salesforce / Tableau |
| Salesforce Agentforce | Salesforce |
| MuleSoft (if applicable) | Salesforce |
Sparx Services handles the implementation service. Software licensing is a direct commercial relationship between your organization and the respective vendors. We provide the BOM so you know exactly what to buy; your procurement team handles the purchase.
Is EA GraphLink different for the Salesforce path vs the Microsoft path?
No. EA GraphLink is the same product serving both paths. It exposes two interfaces: GraphQL and MCP: and both interfaces serve Microsoft and Salesforce tools equally. The same EA GraphLink deployment that connects to Power BI and Copilot can connect to Tableau and Agentforce simultaneously. The Salesforce path isn’t a separate product or a special configuration.
Do I need to deploy Tableau AND Agentforce or can I choose one?
You can choose one. Tableau and Agentforce serve different use cases and different users. Tableau is for BI dashboard consumers: people who want visualization and reporting of architecture data. Agentforce is for AI assistant users: people who want to ask natural language questions in Salesforce. There’s no dependency between them. Organizations often start with one and add the other in a subsequent phase based on showed value.
Is the MCP interface for Agentforce technically different from the Copilot interface?
No. MCP (the Model Context Protocol) is an open standard. EA GraphLink implements a single MCP Server. Microsoft Copilot connects to it via MCP. Salesforce Agentforce connects to it via MCP. The protocol is identical: the distinction is which tool acts as the MCP client. This means the same EA GraphLink MCP Server serves both Copilot and Agentforce connections. Organizations with both M365 and Salesforce users can connect both simultaneously from one deployment.
Can I run both Microsoft and Salesforce integrations from the same EA GraphLink deployment?
Yes. This is one of EA GraphLink’s key characteristics. A single deployment can support Power BI (GraphQL), Microsoft Copilot (MCP), Tableau (GraphQL), and Salesforce Agentforce (MCP) simultaneously. Organizations that span both Microsoft and Salesforce ecosystems don’t need separate EA GraphLink deployments for each. The licensing, configuration, and infrastructure is unified; the integrations are additive.
Does Sparx Services have experience with Salesforce ecosystem deployments?
Yes. The Salesforce ecosystem path: Tableau, Agentforce, MuleSoft: is part of our Connect service offering. We have worked through the Salesforce integration configuration, connector setup, and governance requirements. The Salesforce path is not a secondary or experimental capability: it’s a fully supported delivery track within Connect.
What is the typical Connect engagement scope for the Salesforce path?
Connect for the Salesforce path typically includes: EA GraphLink Foundation deployment, MDG governance validation (or assumes prior Discover completion), and one or more Salesforce integrations (Tableau, Agentforce, or MuleSoft). Scope and investment depend on which integrations are included and the complexity of your environment. Connect ranges from $50K to $185K+, with Salesforce ecosystem scopes typically in the middle of that range for Tableau + Agentforce. Sparx Services scopes Connect specifically for your environment during discovery.
For detailed coverage of each Salesforce ecosystem integration:
For comparison with the Microsoft ecosystem path:
If your organization is Salesforce-centric and you’ve been watching Microsoft-centric EA integration content without a clear path that applies to you, the Salesforce ecosystem path is it. Same foundation. Same interfaces. Different ecosystem.
The starting point is EA GraphLink Foundation: the prerequisite for all integrations. If your MDG governance readiness is uncertain, Discover comes first.
[Start Your Connect Engagement]
Questions about the Salesforce ecosystem path and how it applies to your environment?
[Let’s Discuss Your Integration Options]
Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.