AI Integrations

Sparx EA AI Integrations: Side-by-Side Comparison

Direct Answer

This page is for organizations that know they want to connect Sparx EA to AI and BI tools but aren’t certain which integration: or combination of integrations: fits their ecosystem and use case.

EA GraphLink is the foundation for all integrations covered here. It exposes two interfaces: a GraphQL API (consumed by BI platforms like Power BI and Tableau) and an MCP Server (consumed by AI assistant tools). One EA GraphLink deployment can support multiple integrations simultaneously.

The comparison below covers all 14 integrations across both interfaces. Use it to filter by ecosystem, data residency requirement, primary user, and setup complexity. Then use the decision questions at the bottom to confirm your selection.


The Full Integration Comparison

Integration Primary User Technical Path Ecosystem Data Stays in Env? Prerequisites Beyond EA GraphLink Setup Complexity Best For
Power BI IT Executive, Arch Manager GraphQL Microsoft Yes: Power BI tenant Power BI Pro or Premium license Low Live BI dashboards; Microsoft ecosystem
Microsoft Fabric Data Architect MCP Microsoft Yes: Azure tenant Fabric capacity license High Enterprise data integration; Microsoft data estate
Microsoft Copilot Business Stakeholder MCP Microsoft No: Copilot cloud M365 Copilot license ($30/user/mo) Medium Stakeholder self-service inside M365
Tableau IT Executive, Arch Manager GraphQL Salesforce Yes: Tableau tenant Tableau Creator or Explorer license Low Live BI dashboards; Salesforce ecosystem
MuleSoft Fabric Integration Architect MCP Salesforce Yes: MuleSoft infrastructure MuleSoft subscription High Enterprise data integration; Salesforce ecosystem
Salesforce Agentforce Business Stakeholder MCP Salesforce No: Agentforce cloud Agentforce license Medium Stakeholder self-service inside Salesforce
Kernaro AI Hub Business Stakeholder Via EA GraphLink None: any browser Configurable Kernaro AI Hub license (Sparx Systems) Medium Purpose-built architecture intelligence; any organization
Claude EA Architect, Arch Manager MCP None: universal No: Anthropic API Anthropic API key Low Deep analysis, documentation, complex reasoning
Cursor / Claude Code Systems Engineer, Technical Arch MCP None: universal No: Anthropic API Cursor or Claude Code + API key Low IDE-integrated architecture context for developers
Google Gemini Business Stakeholder MCP Google No: Google cloud Google Workspace Gemini license Medium Stakeholder self-service in Google Workspace
ChatGPT Enterprise EA Architect MCP None: universal No: OpenAI API ChatGPT Enterprise subscription Low Organizations standardized on OpenAI
Azure OpenAI Any user in regulated org MCP Microsoft: Azure Yes: Azure tenant Azure subscription + Azure OpenAI resource High Regulated industries; compliance boundary required
Kernaro Assist EA Architect In-EA extension None: EA client Yes: in EA client Kernaro Assist license (Beta) Low In-EA AI for modeling productivity
AI Modeling Workbench EA Architect Native EA feature None: configurable Configurable Compatible EA version + AI provider credentials Low Evaluation phases; multi-provider flexibility; on-premises AI

EA GraphLink is the foundation: not listed as a comparison row but required for all integrations except AI Modeling Workbench and Kernaro Assist.


How to Choose: by Organization Type

Microsoft-centric organization

Start with Power BI for dashboard connectivity: it uses EA GraphLink’s GraphQL interface and is the lowest-friction integration available. Add Microsoft Copilot for stakeholder self-service if your users are active in M365 and the Copilot licensing is justified. Add Microsoft Fabric only if EA repository data needs to flow into a broader Azure data estate and Fabric is already your data integration platform.

Salesforce-centric organization

Parallel path: Tableau for dashboards (same GraphQL interface as Power BI, Salesforce ecosystem), Agentforce for AI assistant access (MCP, equivalent to Copilot for Salesforce users), MuleSoft Fabric for data integration if it’s already the enterprise integration standard. Start with Tableau and Agentforce; add MuleSoft only if the integration use case is a confirmed priority.

Google Workspace organization

Gemini is the natural AI assistant path. No Power BI or Tableau equivalent within Google’s productivity tools: for BI dashboards, Tableau or Power BI both work regardless of whether the organization uses Google Workspace. Kernaro AI Hub is ecosystem-neutral and provides dedicated architecture intelligence to any browser user.

Mixed or still evaluating

Start with EA GraphLink Foundation: it’s the prerequisite for everything else and establishes the infrastructure. Add Kernaro AI Hub as the first integration: it’s ecosystem-neutral, requires no M365 or Salesforce commitment, and gives stakeholders immediate architecture intelligence access through any browser. This approach keeps options open while delivering value quickly. The AI Modeling Workbench is an alternative evaluation path for organizations that want to test AI providers against their actual architecture tasks before committing.

Regulated industry (defense, government, healthcare, financial services)

Azure OpenAI for AI reasoning: data stays in Azure tenant throughout. Power BI or Tableau for BI dashboards: data stays in the BI platform tenant. Avoid Copilot, Agentforce, Gemini, Claude, and ChatGPT Enterprise unless your compliance review confirms those data flows are permissible. Kernaro AI Hub’s data handling is configurable: confirm the specific deployment model with Sparx Systems for regulated-environment suitability.

Technical and engineering teams

Cursor or Claude Code for IDE-integrated architecture context: architects and developers working in code environments can query the EA repository without leaving their development environment. Claude for deep analysis and complex reasoning tasks. These are the integrations optimized for technical users rather than business stakeholders.

Architects in EA (architect use)

Kernaro Assist for in-EA AI assistance: purpose-built for architecture modeling productivity, available within the EA client. Note: Kernaro Assist is currently in Beta. The AI Modeling Workbench for evaluation and multi-provider flexibility. EA GraphLink + Claude for deep analysis tasks that benefit from frontier model reasoning.


Five Decision Questions

1. Where do our stakeholders spend their time?

This determines which AI assistant integration has the most value. Stakeholders in M365 → Copilot. Stakeholders in Salesforce → Agentforce. Stakeholders in Google Workspace → Gemini. Stakeholders distributed across platforms or outside specific ecosystems → Kernaro AI Hub (any browser, no ecosystem dependency).

2. What is our enterprise BI platform?

Power BI or Tableau are the two dashboard integration options. Both use EA GraphLink’s GraphQL interface and provide live, auto-refreshing dashboards of architecture data. The choice is determined by which BI platform your organization already uses: there’s no meaningful technical reason to prefer one over the other for EA integration specifically.

3. Do we have regulatory constraints on where data can flow?

If yes: specifically if architecture data cannot leave a controlled environment: Azure OpenAI is the correct AI reasoning path. Kernaro AI Hub’s configurable data handling may also be suitable depending on deployment model. All other AI assistant integrations send data to external API infrastructure. If the answer is unclear, a governance review should precede integration planning.

4. Is our primary need stakeholder self-service or architect productivity?

Stakeholder self-service (business executives, program managers, compliance officers querying the repository without architect involvement) → Connect engagement, AI assistant or Kernaro AI Hub integration. Architect productivity (EA architects getting AI assistance during modeling, analysis, and documentation work) → Amplify engagement, Kernaro Assist, or Claude integration. Many organizations need both: but prioritizing one path determines where to start.

5. What is our MDG governance maturity?

All AI assistant integrations surface repository quality directly to users. If your MDG Technology definition is weak: stereotypes inconsistently applied, tagged values poorly populated, relationships incomplete: AI integrations will produce unreliable answers and damage stakeholder trust in the repository. If MDG maturity is uncertain, begin with Discover before planning any AI integration. Discover assesses MDG readiness and develops the governance improvements needed before Connect is viable.


Getting Started

If your MDG readiness is unclear: Start with Discover: the MDG governance assessment that determines whether your repository is ready for AI integration.

If you’re ready to choose and deploy an integration: Connect is the engagement that deploys EA GraphLink Foundation plus one or more integrations.

Individual integration guides:

[Start Your Connect Engagement]

Ready to make your EA investment work harder?

Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.