Government and public sector organizations use Sparx EA for IT modernization programs, agency capability planning, FEAF/TOGAF governance, and DoDAF-aligned acquisition support. The AI integration story in government contexts is different from commercial: data residency requirements typically mandate Azure OpenAI over public AI APIs, and FedRAMP compliance affects how EA data can flow. Repositories accumulated across decades of contractor contributions present a consistent MDG governance challenge that must be resolved before AI integration delivers value.
Key Takeaways
FEAF organizes federal EA across five Sub-architecture Domains: Business, Data, Applications, Infrastructure, and Security. In Sparx EA, these map naturally to ArchiMate’s layers and aspects: with the important addition of FEAF’s Consolidated Reference Models (BRM, DRM, TRM, SRM) as the vocabulary for cross-agency communication. Agencies using the Common Approach to Federal Enterprise Architecture use FEAF as both the governance structure and the communication standard for OMB reporting.
Sparx EA supports FEAF natively through configurable MDG profiles. The critical practice question is whether your agency’s existing repository actually uses those profiles consistently: or whether the models have drifted into ad hoc structures that look like FEAF but don’t behave like it. That’s what a Discover assessment establishes.
TOGAF’s Architecture Development Method (ADM) provides the process governance layer that FEAF lacks. Many federal programs run both: FEAF for the structural vocabulary, TOGAF ADM for the architecture governance cycle: from Architecture Vision through Migration Planning. Sparx EA’s built-in TOGAF support includes the ADM phase structure, Architecture Content Framework artefacts, and the Architecture Repository concept that maps to how federal programs manage architecture versions.
Programs touching DoD components, defense acquisitions, or national security use DoDAF 2.0 as the framework requirement: often alongside or in place of FEAF. Sparx EA’s DoDAF 2.0 support includes the full suite of viewpoints: AV (All Views), OV (Operational), SV (Systems), SvcV (Services), DIV (Data and Information), and StdV (Standards). For programs that cross the civil/defense boundary, Sparx EA supports both frameworks in a single repository with cross-view traceability.
State government EA programs typically align to NASCIO (National Association of State CIOs) guidance rather than FEAF directly. The frameworks are analogous: capability-based planning, IT portfolio management, cloud adoption governance: but the legislative mandates and funding structures differ. Local government EA is less standardized, but the use cases (digital services delivery, legacy system rationalization, shared services architecture) are consistent. Sparx EA’s framework flexibility supports all of these through configurable MDG profiles.
Federal agencies carry application portfolios accumulated over decades: OMB’s annual FITARA scorecard has made this a public accountability issue. Rationalization requires a current-state application inventory mapped to business capabilities, followed by a structured evaluation of each application against TIME model categories (Tolerate, Invest, Migrate, Eliminate). Sparx EA is the repository for both the inventory and the evaluation framework. The AI integration opportunity here is significant: with EA GraphLink connecting the repository to a Kernaro AI Hub query interface, portfolio questions that previously required custom reports become natural language queries: “Which applications supporting the Financial Management capability are candidates for consolidation?”
Cloud migration in government is governed by the Cloud Smart strategy and OMB M-19-17. Architecture work involves documenting the current state application and infrastructure landscape, developing the target state architecture (typically Azure Government Cloud for sensitive workloads, commercial Azure or AWS GovCloud for less sensitive), and building the migration roadmap with dependency sequencing. Sparx EA maintains the full AS-IS to TO-BE progression with dependency tracking across the application and infrastructure layers.
Large federal IT programs: major modernization initiatives, shared services platforms, interagency systems: develop their own program-level architectures that must connect to the agency enterprise architecture. In practice, these often exist as separate repositories that drift apart. Sparx EA’s repository structure and Pro Cloud Server’s multi-repository capabilities support a federated model: program architecture instances linked to the enterprise repository with controlled integration points. Traceability from enterprise capability to program requirement to system specification is the governance mechanism that makes this work.
Federal data sharing mandates: from the Evidence Act to agency-specific data sharing agreements: require documented data architecture that identifies what data exists, where it lives, who owns it, and what sharing arrangements are in place. Sparx EA’s data modeling capability (UML class diagrams, ArchiMate data layer) supports this, and MDG profiles for data classification and sensitivity marking extend it to the classification requirements specific to government data.
FedRAMP authorization is the compliance boundary for cloud services used by federal agencies. As of 2025, Azure OpenAI Service has FedRAMP High authorization: making it the primary option for AI integration that keeps EA repository data within the compliance boundary. Commercial AI APIs (OpenAI’s public API, Anthropic’s API) are not FedRAMP-authorized and cannot be used with repositories containing Controlled Unclassified Information (CUI) or agency-sensitive data.
This has direct implications for EA GraphLink deployment. EA GraphLink transforms the Sparx EA repository via MDG Technology into AI-accessible data: but the AI endpoint it connects to must meet the agency’s compliance requirements. For most federal civilian agencies, that means Azure OpenAI. For the most sensitive programs, it means an on-premises model.
Azure Government Cloud (azure.us) provides the FedRAMP High authorized environment for Azure OpenAI Service deployment. An EA GraphLink implementation in a government context routes through the agency’s Azure Government tenant: the EA repository data never leaves the authorization boundary. This requires an existing Azure Government tenancy with Azure OpenAI Service provisioned, which is a key input to the Discover assessment.
For environments where even Azure Government Cloud introduces unacceptable risk: classified networks, air-gapped systems, JWICS-adjacent programs: Sparx EA’s AI Modeling Workbench supports configuration with locally deployed language models. Open-weight models (Llama, Mistral variants) can be deployed within the classified enclave and connected to the EA repository without any data leaving the boundary. Performance is lower than commercial frontier models, but the use cases most valuable in these environments (requirements traceability queries, model consistency checks, architecture documentation generation) are within the capability of current local models.
A government-scoped Discover engagement assesses:
Government EA programs almost universally inherit repositories built by multiple contractors over many years. Each contractor brought their own modeling conventions, their own MDG profiles (or none), and their own interpretation of whatever framework the program nominally runs. The result is a repository that looks populated but doesn’t support the queries, reports, or analysis that the enterprise architecture is supposed to enable.
Inconsistent MDG governance in government repositories typically manifests as: duplicate elements with different names, relationship types used inconsistently, capability hierarchies that don’t connect to application or technology layers, and tagged values that were defined but never populated. AI integration makes this worse: an LLM querying an inconsistent repository returns inconsistent answers, which is worse than no answer.
Discover addresses this directly with a structured MDG quality assessment. The output is a baseline score, a prioritized list of governance gaps, and a remediation roadmap that sequences MDG cleanup against the planned AI integration program.
What EA frameworks are standard in US federal government?
FEAF (Federal Enterprise Architecture Framework) is the standard for civilian federal agencies, governed by OMB. TOGAF is widely used alongside FEAF for its ADM process structure. DoDAF 2.0 is required for programs with defense and national security components. State governments typically follow NASCIO guidance, which is analogous to FEAF but adapted for state legislative and funding structures.
Does Sparx EA support FEAF?
Yes. Sparx EA supports FEAF through configurable MDG profiles that implement the Sub-architecture Domains and Reference Models. The ArchiMate notation maps cleanly to FEAF’s Business, Data, Applications, Infrastructure, and Security domains. Additional MDG customization can align Sparx EA models to FEAF’s Consolidated Reference Model vocabulary for OMB reporting purposes.
Can Sparx EA be deployed in FedRAMP-authorized cloud environments?
Sparx EA’s Pro Cloud Server can be deployed on Azure Government Cloud (or AWS GovCloud), with the database tier running on FedRAMP-authorized managed database services. The Sparx EA client accesses the repository via the Pro Cloud Server API. This architecture keeps all repository data within the FedRAMP boundary. Sparx Services has experience scoping these deployments as part of a Deploy engagement.
What AI tools are appropriate for government EA repositories?
For repositories containing CUI or agency-sensitive data: Azure OpenAI Service on Azure Government Cloud (FedRAMP High authorized) or on-premises locally deployed models. For repositories confirmed to contain only unclassified, non-sensitive data: commercial Azure OpenAI in a standard Azure tenant is a lower-friction option. Public AI APIs without data processing agreements are not appropriate for any government EA repository.
How do I connect Sparx EA to Azure OpenAI for a government agency?
The connection runs through EA GraphLink, which transforms the Sparx EA repository via MDG Technology into a structured data layer that Azure OpenAI can query. The integration requires: a Pro Cloud Server deployment accessible within the Azure Government network boundary, an Azure OpenAI Service instance provisioned in the agency’s Azure Government tenant, and EA GraphLink configured to route queries through that endpoint. A Connect engagement scopes and implements this architecture.
What is the typical Discover scope for a government agency?
A government Discover engagement typically runs 6–10 weeks and covers: MDG baseline assessment across the full repository, framework compliance review (FEAF/TOGAF/DoDAF as applicable), data classification audit for AI integration planning, Azure Government tenancy review, application portfolio assessment against the existing capability model, and a prioritized recommendations report. Pricing: $25K–$75K depending on repository size and complexity.
How does ITAR affect EA repository AI integration?
ITAR (International Traffic in Arms Regulations) applies to defense articles and related technical data. If the EA repository contains ITAR-controlled technical data: system specifications, interface definitions, or capability descriptions for defense articles: that data cannot be exported or transmitted to non-US persons or unapproved cloud environments. This restricts AI integration options significantly: Azure Government Cloud with proper data handling agreements may be acceptable; most commercial cloud AI APIs are not. On-premises deployment is the safest path for ITAR-adjacent repositories.
What is the difference between agency EA and program architecture in Sparx EA?
Agency EA captures the enterprise-level view: business capabilities, the application portfolio supporting those capabilities, the infrastructure layer, and the strategic roadmap. Program architecture is a project-scoped view developed for a specific IT modernization or acquisition program: focused on the specific systems, interfaces, and capabilities the program delivers. In Sparx EA, these typically live in separate model packages or separate repository instances, linked by shared elements (capability references, technology standards). Pro Cloud Server’s multi-repository capabilities support federated governance across agency and program architectures.
Government EA programs need architects who understand FEAF, FedRAMP, and the contractor-contribution governance problem: not a generic EA consulting pitch.
Discover ($25K–$75K): MDG readiness assessment, maturity baseline, AI integration feasibility scoped to your agency’s framework requirements and compliance environment.
Platform Support: Ongoing Sparx EA technical expertise for government programs, including Pro Cloud Server deployment, MDG governance, and DoDAF/FEAF framework configuration.
Start a Discover Conversation → | DoDAF Practice Guide → | Azure OpenAI Integration Guide →
Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.