SEE IT IN ACTION
The Challenge
The rationalization exercise starts with data collection, not analysis. Lifecycle values to normalize. Ownership fields to chase. Platform names to standardize. By the time the analysis can begin, you have spent two weeks compensating for gaps that were already in the model.
The categorization survives in the workshops because the people who built it are in the room. At the executive level, a CFO who was not in the workshop asks a question the slides cannot answer. The recommendation gets deferred. The follow-up takes a week.
"What depends on the top retirement candidate?" It is a reasonable question. The answer is in the model. Getting it out takes a week. The pause during the presentation is the moment the board's confidence in the process shifts.
When you use AI Augmented Architecture for Application Portfolio Management
You walk into the first rationalization session with the current model view already in hand. Open Claude alongside Sparx EA. Run the portfolio profile. The view by lifecycle, criticality, business owner, and technology platform comes back in one session. Data quality gaps appear as part of the output, organized by how much they affect rationalization confidence, so you address the gaps that matter before presenting.
End-of-life lifecycle status. Deprecated platform with no upgrade path. No active downstream consumers. Functional overlap with a named alternative. The rationale is in plain language. The evidence is verifiable in the model. Each recommendation can be examined, questioned, and defended from the repository.
Applications with no documented inbound or outbound connections are flagged. Applications where the connectivity data looks incomplete are flagged separately with a recommendation to verify, not automatically classified as orphans. The output distinguishes confirmed orphans from data quality gaps.
When the board asks about dependencies during the presentation, you type the question into Claude. The dependency graph for any retirement candidate comes back in 30 seconds from the current model. Decision makers trust an answer from the model differently than an answer from recall. The architect's credibility is enhanced.
AI Power Tools for EA surfaces the candidates and the evidence. The analysis makes the decisions defensible. It does not make the decisions. Every retire, modernize, and retain outcome is a business and architectural judgment made by your leadership.
UNDER THE HOOD
AI Power Tools for EA connects to a running Sparx EA session on your computer through EA's automation interface. It reads the repository, tagged values, and connector topology. It runs through Claude. If you have related data in other systems — such as your CMDB — that data can be read and analyzed alongside your EA model.
For application portfolio management, the relevant tools are listed below. Each tool maps to a specific question in the rationalization workflow.
| aggregate_portfolio | Builds the portfolio profile by lifecycle, criticality, business owner, and technology platform. |
| summarize_tagged_value_usage | Surfaces tagged-value coverage gaps organized by rationalization relevance. |
| find_orphan_elements | Identifies applications with no documented connections. |
| traverse_element_subgraph | Walks the dependency graph in response to questions like "what depends on this system?" |
| get_element_business_view | Produces the plain-language rationale for each rationalization candidate. |
| export_diagram_image | Exports the current diagram view for inclusion in the board deck. |
FOR TEAMS ALREADY ON SPARX EA
If you are already running APM in Sparx EA, AI Power Tools for EA is how you do it better. The data collection sprint shrinks from weeks to days. The rationalization candidates arrive with evidence the board can examine. The board presentation survives live questions.
If you are not yet running APM systematically in Sparx EA, this is how you start. The portfolio profile from the existing model gives you the current state before you invest in a dedicated APM process. You build from what you already have.
AI POWER TOOLS FOR ENTERPRISE ARCHITECT
86 tools. Full read/write access to your Sparx EA model. Seven-day free trial. No credit card required to start.
HOW TO GET STARTED
Understand your starting point
Get AI Power Tools for EA running
Build the habit across your team
FREQUENTLY ASKED QUESTIONS
Open your architecture model in EA, then go to Claude and type in a prompt describing what you want to analyze. If you only want to look at a specific business unit, location, or another subset, provide those constraints. The Skills Library that comes with AI Power Tools for EA handles most of the logic. Claude runs the rationalization directly from the EA repository, profiling applications by lifecycle, criticality, business owner, technology platform, and connectivity. It surfaces retirement and modernization candidates with the rationale in plain language. Data quality gaps appear as part of the output, organized by how much they affect rationalization confidence.
Not necessarily. If your EA model has the application inventory, connectivity data, and lifecycle and ownership tagged values populated to a reasonable standard, AI Power Tools for EA can run the rationalization analysis from that existing repository. If you have a separate APM tool, you can connect that to Claude as well to provide real-time access to the source data. The question is whether the incremental value of a second tool justifies a second data maintenance process. Organizations whose primary challenge is rationalizing the portfolio often get full value from the EA model they already have.
The repository does not need to be clean before analysis begins, but the quality of the rationalization is highly dependent on the data you have available. AI Power Tools for EA profiles data quality as part of the rationalization output. Applications with incomplete ownership, missing lifecycle status, or unvalidated connectivity are flagged by confidence level. High-confidence rationalization candidates are those where the data supports the recommendation. Low-confidence candidates are surfaced with the specific gap noted. You address the critical gaps before presenting to the board without running a full cleanup sprint first.
No. AI Power Tools for EA surfaces rationalization candidates with the evidence and rationale from the model. The retire/modernize/retain decisions are business and technical decisions made by your organization's leadership. AI should augment and support the decisions but the final call is a human one. What changes is that the evidence arrives in plain language and is traceable to the model, so the decisions can be made with confidence and defended during board scrutiny.
With AI Power Tools for EA running alongside Sparx EA, you type the question into Claude during the presentation. The traverse_element_subgraph tool walks the dependency path from the application in question and returns the result in plain language within 30 seconds. The answer comes from the current model, not from a spreadsheet prepared weeks earlier.
The first conversation covers what state your APM data is in, what the engagement pattern looks like, and whether the model you already have is ready to support a rationalization analysis. Bring your EA model state and your rationalization cycle timeline.
Schedule a Discovery Call