Connect
Microsoft is the default enterprise platform for most large organizations. Power BI is the dominant business intelligence tool. Microsoft Copilot is the enterprise AI assistant, embedded in Teams, Outlook, and Microsoft 365. Microsoft Fabric is the unified analytics platform. And Sparx EA is the enterprise architecture repository.
In 2026, all of these can be connected: from a single EA GraphLink deployment. The result is a Microsoft-native EA intelligence platform: architecture data flowing from the Sparx EA repository into Power BI dashboards, into Microsoft Fabric data pipelines, and into Microsoft Copilot conversations: all within the organization’s existing Microsoft tenant.
This is not a future roadmap. It is an available architecture today. Sparx Services implements it through a Connect engagement.
EA GraphLink is the connectivity product developed by Sparx Systems: the makers of Sparx EA: that transforms the Sparx EA repository into a structured data source accessible to external systems.
EA GraphLink operates through two interfaces:
Interface A: GraphQL API: A structured query API that exposes the EA repository as a graph of elements, relationships, and properties. External systems can query this API to retrieve architecture data in structured form. Power BI is the primary consumer of Interface A.
Interface B: MCP Server: A Model Context Protocol server that makes EA repository data available to AI language models and agentic platforms. Any AI system that supports MCP can connect to an EA GraphLink MCP server and query the repository in natural language, receiving structured context drawn from the live EA data. Microsoft Copilot, Microsoft Fabric Agents, Kernaro AI Hub, and other MCP-compatible AI tools are consumers of Interface B.
One EA GraphLink deployment provides both interfaces simultaneously. Deploying EA GraphLink for Power BI integration automatically enables the MCP server for AI integration: organizations do not need two separate deployments for two separate purposes.
EA GraphLink enables connection to three distinct Microsoft products. Each does something fundamentally different. It is important not to conflate them.
What Power BI does with EA data: Power BI consumes structured data from EA GraphLink’s GraphQL API and renders it as visual dashboards: charts, matrices, heat maps, tables, and composite portfolio views. These dashboards refresh automatically as the EA repository is updated.
What this looks like in practice: An executive opens a Power BI dashboard labelled “Architecture Portfolio” in their browser. They see a capability heat map: 40 capabilities color-coded by maturity level, with investment direction indicators. They click on a capability and see the applications that support it, the technology components those applications run on, and the projects currently modifying that landscape. They do not touch Sparx EA. They do not need an EA tool license. The dashboard is live, governed data from the EA repository.
The semantic model: EA GraphLink exposes a semantic model: a structured representation of the EA repository’s elements and relationships: that Power BI can use for ad hoc reporting. This is not a static export; it is a live connection. When an architect updates an application’s status in Sparx EA, the Power BI dashboard reflects the change on the next refresh.
Typical Power BI dashboards from EA data:
Licensing: Power BI is licensed separately by the client organization from Microsoft. Sparx Services implements the EA GraphLink interface and builds the initial dashboard set. Ongoing Power BI licensing is a Microsoft subscription.
What Microsoft Copilot does with EA data: Microsoft Copilot (the enterprise AI assistant embedded in Teams, Outlook, and Microsoft 365 applications) can be extended with plugins that give it access to external data sources. EA GraphLink’s MCP Server is the mechanism by which Sparx EA repository data becomes available as a Copilot plugin data source.
When connected, a user in Microsoft Teams can type: “Which applications in our portfolio are approaching end of life and are not yet part of a remediation project?”: and Copilot returns a structured answer drawn from the live EA repository, formatted for conversation. This is architecture intelligence in the flow of work: in Teams chats, in Outlook drafts, in Word documents, without requiring the user to open Sparx EA, run a query, or wait for a report.
Specific Copilot capabilities from EA GraphLink:
What Copilot is NOT doing: Microsoft Copilot is not modifying the EA repository, not making architecture decisions, and not operating autonomously. It is a natural language interface to EA repository data: answering questions, surfacing information, and supporting informed decisions. The architecture judgement remains with the architects.
The Microsoft tenant boundary: Microsoft Copilot operates within the organization’s Microsoft 365 tenant. EA data accessed through the MCP plugin does not leave the tenant boundary. For regulated industries: financial services, defense, healthcare: this is a material data governance requirement. EA GraphLink data served to Copilot is processed within Microsoft’s enterprise trust boundary, not sent to an external AI service.
Licensing: Microsoft 365 Copilot requires a Microsoft 365 Copilot license (per user, from Microsoft). The EA GraphLink MCP plugin requires EA GraphLink deployment. Sparx Services implements both the EA GraphLink configuration and the Copilot plugin integration.
What Microsoft Fabric is: Microsoft Fabric is Microsoft’s unified analytics platform: combining data engineering, data warehouse, data science, real-time analytics, and business intelligence in a single SaaS platform. It includes Fabric Lakehouse (for data storage and processing), Fabric Data Factory (for data pipelines), Fabric Notebooks (for data science), and OneLake (the universal data store that underlies all Fabric workloads).
Fabric is Microsoft’s strategic data platform: the successor trajectory from Azure Synapse and the integration point between Power BI, Azure Data Services, and enterprise data estates.
What EA GraphLink enables in Fabric: EA GraphLink’s MCP Server exposes EA repository data as a queryable data source that Microsoft Fabric can consume. This enables:
EA data as a Fabric data source: Fabric Data Factory pipelines can extract architecture data from EA GraphLink’s API and load it into OneLake: making EA repository data available alongside operational data, financial data, and other enterprise datasets in the Fabric data estate. An architect’s capability map can be joined to a financial system’s cost allocation data to produce a capability-cost view that neither system can produce alone.
Cross-source data enrichment: In Fabric, EA data becomes a dimension in multi-source analytics. An application performance dataset (from Azure Monitor) can be enriched with the application’s architecture metadata (from EA GraphLink) to produce a view that correlates technical performance with architectural characteristics. A project portfolio dataset (from a PPM tool) can be enriched with EA impact assessments from the repository.
Fabric Lakehouse with architecture data: When EA data is ingested into a Fabric Lakehouse, it becomes available as a dataset for Fabric Notebooks: enabling data science and ML work on architecture data at scale. Pattern recognition across large EA repositories, predictive modeling of technology risk, and automated gap detection are all tractable in Fabric that are not tractable in the EA tool directly.
Agentic data flows: Microsoft Fabric includes agentic capabilities that can orchestrate multi-step data operations using AI. An EA GraphLink MCP connection enables Fabric agents to query the EA repository as part of multi-step workflows: for example, an agent that monitors for applications approaching end of life (from the EA repository), checks for active remediation projects (from the PPM tool), and generates exception reports for architecture governance review.
The Fabric tenant boundary: Like Copilot, Microsoft Fabric operates within the organization’s Microsoft tenant: data processed in Fabric stays within the Microsoft enterprise trust boundary.
Licensing: Microsoft Fabric requires a Fabric capacity license (from Microsoft). Fabric capacity licensing is per workspace and scales with compute requirements. Sparx Services implements the EA GraphLink connection and the initial Fabric data pipeline configuration.
A recurring concern for regulated industry organizations considering AI-powered EA tooling is data residency and governance. Where does the architecture data go? Who can access it? Does it leave the organization’s control?
The Microsoft ecosystem path answers this concern directly. When EA GraphLink data flows to Power BI, Microsoft Copilot, or Microsoft Fabric, it flows within the organization’s Microsoft 365 tenant. Microsoft’s enterprise trust boundary: including data residency controls, access governance, and compliance certifications: applies to the EA data as it does to any other enterprise data in the tenant.
This matters in specific regulated contexts:
EA data processed by Kernaro AI Hub (connected via EA GraphLink MCP) or by general-purpose AI tools (Claude, ChatGPT Enterprise, Azure OpenAI) operates under different data governance conditions. For organizations where tenant boundary is a hard requirement, the Microsoft ecosystem path is the appropriate choice.
Deploying the full Microsoft EA platform requires the following:
Sparx EA repository: A governed, active Sparx EA repository: ideally with consistent MDG (Model Driven Generation) profiles defining stereotypes, tagged values, and relationship constraints. The quality of Power BI dashboards, Copilot answers, and Fabric data pipelines is bounded by the quality of the underlying repository data. A poorly governed repository will produce low-quality outputs from all three Microsoft integration paths.
Sparx EA Pro Cloud Server (PCS): EA GraphLink requires the Pro Cloud Server, Sparx Systems’ server-side infrastructure for Sparx EA. PCS provides the database backend (typically SQL Server or PostgreSQL), authentication, and the API layer that EA GraphLink connects to.
EA GraphLink license: EA GraphLink is licensed by Sparx Systems, not Sparx Services. Sparx Services will advise on the appropriate EA GraphLink configuration for the organization’s Microsoft integration requirements.
Power BI Premium or Fabric workspace (for Interface A): To use EA GraphLink’s DirectQuery connection to Power BI in a scalable way, a Power BI Premium per capacity license or a Microsoft Fabric workspace is required. Power BI Free and Pro licenses support import mode but not scalable DirectQuery against external APIs. Microsoft licenses this separately.
Microsoft 365 Copilot license (for Interface B → Copilot): Microsoft 365 Copilot is a per-user add-on license on top of Microsoft 365 E3 or E5. The EA GraphLink MCP plugin requires Copilot to be active for the users who will use the EA intelligence capability.
Microsoft Fabric capacity (for Interface B → Fabric): Fabric operates on a capacity-based licensing model. The appropriate Fabric capacity tier depends on the volume of EA data, the complexity of Fabric pipelines, and the number of users accessing Fabric datasets. Microsoft licenses this separately.
MDG quality: The single most important non-technical prerequisite. EA GraphLink exposes what is in the repository. If elements are untyped, tagged values are sparsely populated, or relationships are inconsistent, the Microsoft integration will surface that quality gap: to Power BI users, to Copilot users, and to Fabric pipelines. Sparx Services recommends a Deploy engagement to establish or validate MDG governance before a Connect engagement to deploy the Microsoft integration.
A Connect engagement deploying the Microsoft EA platform includes:
Connect engagements start at $50K and scale to $185K+ depending on the scope of Microsoft integration (one, two, or all three paths), the complexity of the MDG governance work required, and the number of Power BI dashboards and Fabric pipelines in scope.
Does EA GraphLink connect to Power BI, Copilot, and Fabric from a single deployment?
Yes. One EA GraphLink deployment provides Interface A (GraphQL API, consumed by Power BI) and Interface B (MCP Server, consumed by Microsoft Copilot and Microsoft Fabric) simultaneously. Deploying EA GraphLink for Power BI integration automatically activates the MCP Server endpoint. Organizations do not need separate deployments for each Microsoft product they want to connect.
Who licenses Power BI, Microsoft Copilot, and Microsoft Fabric?
Microsoft licenses Power BI, Microsoft 365 Copilot, and Microsoft Fabric: they are Microsoft products. EA GraphLink is licensed by Sparx Systems. Sparx Services is the implementation partner that deploys EA GraphLink, configures the integrations, and builds the initial Power BI dashboards, Copilot plugins, and Fabric pipelines. The ongoing Microsoft product subscriptions are the client’s responsibility as part of their Microsoft licensing agreement.
What is the minimum Sparx EA repository quality needed for useful integration outputs?
There is no fixed threshold, but a practical standard is: elements should have consistent stereotypes from a governed MDG profile, key tagged values (owner, maturity, status, lifecycle) should be populated on strategic elements, and relationships between capability, application, and technology elements should be consistently modelled. A repository that lacks these basics will produce low-quality Power BI dashboards, unreliable Copilot answers, and sparse Fabric datasets. Sparx Services assesses repository readiness as part of Connect engagement scoping.
Does EA data leave the Microsoft tenant when Copilot or Fabric processes it?
No. Microsoft Copilot and Microsoft Fabric both operate within the organization’s Microsoft 365 tenant. EA data accessed via EA GraphLink’s MCP Server and processed by Copilot or Fabric is handled within Microsoft’s enterprise trust boundary. It is subject to the organization’s Microsoft tenant data residency configuration, access controls, and compliance certifications. This is a material consideration for regulated industries: financial services, defense, healthcare: where architecture data governance is a compliance requirement.
What is the difference between what Power BI, Copilot, and Fabric do with EA data?
They do three fundamentally different things. Power BI visualises EA data: it turns repository content into dashboards, heat maps, and portfolio views for executive and management audiences. Microsoft Copilot makes EA data conversationally accessible: it answers natural language questions about the architecture from within Teams, Outlook, and other Microsoft 365 applications. Microsoft Fabric analytically processes EA data: it ingests, enriches, and models architecture data alongside operational, financial, and project data for complex multi-source analytics. All three are enabled by a single EA GraphLink deployment.
Can we use EA GraphLink with Copilot without also using Power BI?
Yes. EA GraphLink’s Interface A (GraphQL → Power BI) and Interface B (MCP → Copilot/Fabric) are independent. An organization can deploy EA GraphLink exclusively for Copilot integration without building Power BI dashboards. In practice, most organizations find that the Power BI integration delivers immediate visible value to a wider audience, while the Copilot integration delivers more sophisticated value to users who actively query the architecture. Sparx Services recommends starting with Power BI (Interface A) as it is easier to show ROI to stakeholders.
What is the Pro Cloud Server and is it required?
Sparx EA Pro Cloud Server (PCS) is Sparx Systems’ server infrastructure for Sparx EA: it provides the database backend (SQL Server or PostgreSQL), web-based EA access, API services, and the platform that EA GraphLink runs on. EA GraphLink requires PCS; it cannot connect directly to a file-based (.eapx) EA repository. Most enterprise Sparx EA deployments already use PCS. If the organization is using a file-based repository, Sparx Services will scope the PCS migration as part of a Connect engagement.
How long does a Connect engagement take?
A Connect engagement deploying the Microsoft EA platform typically takes 8–16 weeks from scoping to initial go-live, depending on the repository readiness, the scope of Microsoft integrations, and internal IT governance requirements (firewall changes, API registration in Azure AD, Power BI workspace provisioning). Repository governance remediation, if required, adds time. Sparx Services structures Connect engagements in phases: EA GraphLink and Power BI first, then Copilot plugin, then Fabric: so value is delivered progressively rather than all at program end.
If your organization runs Microsoft 365: Power BI, Teams, Fabric: and Sparx EA, you have the foundations for an architecture intelligence platform that operates within your existing Microsoft tenant.
A Connect engagement delivers the full EA GraphLink deployment, Power BI dashboards, Copilot plugin integration, and (where in scope) Microsoft Fabric pipeline configuration.
Talk to Sparx Services about a Connect engagement
Connect engagements start at $50K. Scoped after an initial repository readiness assessment.
Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.