AI Power Tools for EA
Your architects already use Sparx EA. The AI augmented architecture practice your organization needs does not require migrating away from it. AI Power Tools for EA connects AI to your live repository and delivers AI-assisted modeling, analysis, governance, and stakeholder engagement against your own models, your own standards, and your own MDG. Works with Claude Cowork, Claude Code, and GitHub Copilot in VS Code today. M365 Copilot support is coming soon.
Add the full capabilities of Claude AI to the Sparx Enterprise Architect you already know
AI Power Tools for EA runs alongside Sparx EA on your architect's computer. Sparx EA stays open. The architect stays in charge. The AI joins the session as a capable collaborator who takes direction, works the repository, and handles the mechanical steps that currently consume the most time.
The architect provides the intent and the instructions. What needs to be modeled. What standards to check against. Who needs to be briefed and at what level. The AI handles the execution: navigating the repository, creating and linking elements, running conformance checks, assembling diagrams, drafting deliverables. The judgment stays with the architect. The mechanical heavy lifting moves to the AI.
Nothing gets replaced. Your repository stays where it is. Your MDG-defined standards stay in place. Your architects keep working in the same Sparx EA they already know. You add the AI capabilities of Claude to the tool your team already uses and trusts.
Runs alongside Sparx EA
Keep Sparx EA open. The EA MCP Server connects to the running instance via COM automation and gives the AI full access to your live repository: elements, packages, diagrams, connectors, tagged values, and metadata. No exports. No copy-paste. The AI reads and writes directly to the model your architects are working in right now.
Architect drives, AI executes
Tell the AI what you need: "Model this system from the attached spreadsheet," "Check this package for violations against our standards," "Produce a briefing for the steering committee." The AI handles the steps. The element creation, the stereotype assignment, the conformance pass, the document assembly. Your architectural judgment stays in the loop. The repetitive work moves out of it.
Your choice of AI client
The same toolchain and the same EA connection are available from three AI interfaces today. Claude Cowork is the designed-for-this experience, built for architects who want guided AI workflows without a coding environment. Claude Code suits power users and engineers who work from the terminal. GitHub Copilot in VS Code brings the same capabilities to teams already standardized on GitHub. M365 Copilot support is coming.
Four use case domains, one connected toolchain
AI Power Tools delivers across all four dimensions of architecture work. Each domain has its own audience, its own output shape, and its own set of scenario-specific skills. All four run through the same connected EA repository, whichever AI client you use to access them.
Architecture modeling
Creating, reverse-engineering, maintaining, and standardizing architecture models. Architects turn spreadsheets, documents, and inventories directly into properly stereotyped, fully connected EA diagrams. MDG management works across all Sparx-supported modeling languages and client-built standards.
- Turn a requirements document or spreadsheet into a complete EA diagram in minutes
- Reverse engineer cloud infrastructure, code, or CMDB data into stereotyped models
- Build, extend, and validate custom MDG technology files without hand-editing XML
- Ingest requirements from Jira, Excel, or Word with live traceability links to design elements
- Assess and classify your repository's MDG situation automatically at engagement kickoff
Architecture analysis
Extracting technical insight from the architecture repository: dependency analysis, impact assessment, relationship discovery, and ad-hoc queries. The anchor capability is relationship discovery. Architects compare two sets of objects, adjudicate candidate connections, and persist accepted relationships directly in EA.
- Discover relationships between EA elements and external systems including Jira and regulatory sources
- Trace the full downstream impact of changing or retiring any application or component
- Analyze dependency health: orphans, fan-in/fan-out scores, cyclic dependency detection
- Identify portfolio redundancies and consolidation candidates with evidence-backed rationale
- Run ad-hoc repository queries in plain language, no SQL or EA schema knowledge required
Architecture governance
Enforcing standards, auditing conformance, and maintaining platform health across both the modeling content and the EA platform infrastructure. Rules live in MDG XML and a companion YAML sidecar, keeping the AI grounded in your organization's actual standards rather than generic interpretations.
- Validate models against your organization's governance rules as architects work, not at review time
- Run formal conformance audits with executive-ready findings and a remediation roadmap
- Generate machine-executable governance rules from a written modeling standards document
- Audit EA user access, Prolaborate permissions, and deployed platform assets
- Track repository health over time: ownership coverage, orphan rates, convention drift
Stakeholder engagement
Architecture knowledge is currently locked inside Sparx EA, accessible only to licensed users who know the tool. This use case changes that. Stakeholders ask questions of the model in plain language and get answers grounded in the organization's actual data, without EA access or architect intermediation.
- Answer stakeholder questions about the architecture in plain business language
- Expose the EA repository to M365 Copilot and enterprise AI agents as a live data source
- Generate Power BI dashboards and structured data extracts directly from the repository
- Produce architecture documentation written for business readers, no UML terminology visible
- Let project teams model requirements in EA without EA training or a modeling queue
Give your preferred AI platform the ability to work directly in Sparx EA
The EA MCP Server connects to the Sparx EA application running on your computer — not to an export, not to a snapshot, but to the live session. Your AI platform can create elements, build diagrams, update connectors, and run governance checks the same way you would. And because it is operating the running application directly, you see every change appear in your model as it happens.
Connect from the AI interface that fits how your team already works. Three are available now. A fourth is in progress.
Claude Cowork
Sparx Services designed the toolchain to run natively in Claude Cowork. EA stays open on one side of your screen. Cowork runs alongside it. You give instructions in plain language, and you watch the model update in real time as Claude works. No development environment, no command line, no configuration beyond the initial connection.
Claude Code
Claude Code gives technical architects and power users the same live EA access from the terminal. EA stays open. Changes appear in the running application as each tool call completes. Ideal for engineers extending the skills, running batch operations, or working sessions where step-by-step control matters.
GitHub Copilot in VS Code
Connect the EA MCP Server to GitHub Copilot via VS Code's built-in MCP support. EA runs on your computer. Copilot operates it directly from inside your editor. Architecture work happens alongside the code and configuration it supports, without switching tools or exporting anything.
M365 Copilot
M365 Copilot support will extend live EA access to the tools most stakeholders already use every day. A business leader in Teams will ask a question and get an answer grounded in the running EA model, with no need to open Sparx EA or contact the architecture team.
How the components fit together
The toolchain runs in three coordinated layers: a native EA connector that bridges the AI to your live repository, an AI orchestration layer that interprets intent and composes workflows, and a skills layer that packages those workflows into scenario-specific deliverables. Client interfaces sit above all of this. You access the toolchain from whichever environment fits how your team works.
Eight scenarios, one connected toolchain
AI Power Tools is built around the recurring architecture work that consumes the most time. Each scenario has its own skills, its own output shape, and runs through the same toolchain against your existing Sparx EA repository.
Architects spend more time managing the EA tool than doing architecture work. AI Power Tools handles the repository mechanics: turning project briefs into structured models, assigning stereotypes, building diagrams, and validating conformance against your MDG standards as the work progresses. The architect focuses on decisions. The AI handles the steps.
Build and maintain the MDG technologies and reference models that your practice runs on. AI Power Tools analyzes the existing repository to identify observed patterns, generates or extends MDG technology files without hand-editing XML, and keeps governance rules current as standards evolve. Standards become living artifacts rather than documents that drift.
Architecture review board preparation takes two days every month. AI Power Tools connects to your Sparx EA repository, runs the health scorecard and conformance pass, categorizes every finding into three buckets, and produces the complete reviewer-ready report. The board gets the same evidence base. The steward gets two days back.
Two organizations' architecture repositories rarely speak the same language. AI Power Tools compares applications, capabilities, infrastructure, and processes across both models, resolves naming convention differences semantically, and surfaces consolidation candidates with supporting evidence. Decisions persist as traceable connectors in the combined model.
Translate regulatory obligations into traceable architecture relationships. AI Power Tools queries applicable regulations directly, maps requirements to your EA capability model, and proposes capability-to-regulation connectors for architect review. Accepted mappings build a full audit trail inside the model, not in a separate spreadsheet.
Turn cloud infrastructure into architecture models. AI Power Tools ingests Terraform state, cloud console exports, and CMDB data and produces a stereotyped EA baseline. From that baseline, it analyzes dependency health, traces the impact of retiring or moving each component, and produces sequencing recommendations grounded in the actual dependency graph.
Maintain a living view of the application portfolio inside EA. AI Power Tools ingests spreadsheets and CMDB exports, builds structured portfolio models, analyzes redundancies and lifecycle distribution, and enables business stakeholders to ask portfolio questions in plain language. Portfolio dashboards stay current without a manual extraction cycle.
When an incident fires, the time to identify impact is measured in minutes. AI Power Tools traverses the full dependency graph from the failing component, identifies all affected business processes and services ranked by criticality, and maps impacted systems to organizational units for targeted communications. The same capability serves change impact assessment before every CAB review.
Your EA practice can run on AI augmented workflows now.
Talk to us about a 30-day evaluation engagement scoped to your repository. At the end, you have concrete evidence of the productivity and quality changes — or you don't continue.