AI Integrations
Kernaro AI Hub is a browser-based portal that brings natural language intelligence to your Enterprise Architect repository. Any stakeholder, business executive, program manager, compliance officer, IT strategist, can open a browser, ask a question in plain English, and receive an instant answer from live repository data. No architecture license required. No architect needed to interpret the question or translate the answer. Kernaro AI Hub is published by Sparx Systems and reached general availability in 2026, making it a stable, production-ready platform. It runs on top of EA GraphLink, which transforms the physical EA repository schema into an intelligent, query-ready knowledge graph. The MDG Technology definition in your repository directly affects the quality of Hub’s answers, if your repository governance is strong, Hub’s answers are precise; if governance is weak, Hub reflects that weakness. This is why MDG readiness assessment is the prerequisite step before deployment.
Before Kernaro AI Hub:
After Kernaro AI Hub:
The shift is dramatic: from asynchronous, architect-mediated answers to synchronous, stakeholder self-service intelligence. For large stakeholder bases, this eliminates the single biggest source of architect interrupt work and frees capacity for strategic initiatives.
Kernaro AI Hub excels at questions that require current repository data and don’t need human judgment. Real examples from Sparx Services practices:
Application Lifecycle Management “Which applications are approaching end-of-life in the next 18 months?” Kernaro AI Hub returns a current list grouped by business owner and support status. No spreadsheet needed. No weekly manual status report required.
Capability Mapping “What technology stack supports our customer onboarding capability?” An IT executive gets an instant view of all applications, integrations, and infrastructure supporting that capability. This would traditionally require a manual briefing from an architect.
Change Impact Assessment “What applications and services would be affected if we migrated this database platform?” Instead of a weeks-long impact assessment, Kernaro AI Hub returns a current dependency map in seconds.
Compliance and Risk Coverage “What percentage of our portfolio meets the zero-trust architecture standard?” A compliance officer gets an instant metric instead of asking the architecture team to run a manual audit.
Investment Justification “What is the architecture cost breakdown for the digital transformation program?” A CFO can understand cost distribution across infrastructure, licensing, and integration without architect intervention. This supports better business cases.
Portfolio Rationalization “Show me all applications with deprecated technology in the stack.” A CTO running a modernization initiative gets current inventory without relying on spreadsheets updated quarterly by architects.
Risk and Vendor Management “Which applications depend on this vendor’s platform?” Risk and legal teams get a real-time view when a vendor relationship is at risk, instead of waiting for architecture to respond.
These use cases share a pattern: they require current repository data, are time-sensitive, have a large audience, and occur frequently. Kernaro AI Hub is purpose-built for this workload.
Kernaro AI Hub concentrates value in four stakeholder personas:
| Persona | Why they benefit | Typical ROI |
|---|---|---|
| Business Executive | Immediate access to architecture decisions affecting strategy, without architect bottleneck. Supports faster decision-making. | Self-service answers to 20-30 questions/quarter that would otherwise interrupt architecture team; equivalent cost avoidance at ~$15K-$25K/year per executive |
| Program Manager | Daily architecture questions during program execution. Real-time dependency, risk, and impact data instead of briefing requests. | 2-4 hours/week architect interrupt time eliminated; equivalent to 1-2 FTE reallocation |
| Risk & Compliance Officer | Continuous audit trail and compliance posture visibility. Answers to “What meets X standard?” appear instantly instead of requiring manual assessment. | Audit cycle acceleration; reduced assessment hours; faster policy coverage verification |
| Architecture Manager | Interrupts drop 50-70%; team capacity increases 15-25% for strategic work. Visibility into how stakeholders understand architecture. | 1-2 FTE hours/week recovered for each stakeholder tier using Hub |
| IT Executive / Chief Architect | Portfolio visibility at a glance. Stakeholders self-service rather than escalating to senior leaders. | Depends on escalation volume; typically 5-10 hours/week executive interrupt time eliminated |
Less value for: Architects themselves (who already query EA directly) and organizations with low stakeholder question volume (fewer than 10/week across the board).
Architecture Self-Service at Scale
The central value proposition is eliminating the architect-as-bottleneck pattern. Organizations with distributed stakeholders, large program portfolios, or mature governance patterns accumulate architecture questions faster than architects can answer them. Kernaro AI Hub inverts this dynamic: stakeholders get self-service access to intelligence.
Purpose-Built for Architecture Intelligence
Kernaro AI Hub is not a general-purpose AI assistant like Copilot or Agentforce. It is designed specifically to answer questions about architecture: applications, capabilities, infrastructure, dependencies, standards, compliance. It understands EA’s domain model (elements, relationships, tagged values, MDG stereotypes) natively. This focus means higher accuracy, more relevant answers, and better integration with your EA governance.
No M365 or Salesforce Licensing Dependency
Unlike Microsoft Copilot (which requires M365 licensing) or Salesforce Agentforce (which requires Salesforce licensing), Kernaro AI Hub connects any stakeholder with a browser and network access. Your entire organization can use it regardless of their M365 or Salesforce seat. This matters for organizations where architecture visibility needs to reach constituencies outside standard enterprise tool licensing.
Generally Available, Production-Ready
Kernaro AI Hub reached GA in 2026. It is not experimental. It has a published roadmap, documented API, support model, and upgrade path. This is different from beta products where feature stability and direction are uncertain.
Clear ROI on Interrupt Cost
The financial case is straightforward: measure your current architect interrupt cost (weekly questions × time per question × architect day rate), estimate self-service adoption rate (typically 50-70% of current ad-hoc questions), and the annual hours recovered become clear. For a 5-person architecture team fielding 50 ad-hoc stakeholder questions per week with an average 45-minute response cycle, moving 60% of those questions to self-service recovers roughly 90 hours/quarter (2.5 FTE-weeks), equivalent to $70K-$130K annually in recovered capacity depending on seniority mix.
EA GraphLink Must Be Deployed First
Kernaro AI Hub cannot connect to the raw EA repository schema. It requires EA GraphLink as a prerequisite, the connectivity and data transformation layer that makes the repository queryable. If GraphLink is not yet deployed in your environment, the deployment effort (complexity, timeline, infrastructure) needs to be factored into the Kernaro decision. For organizations already running EA GraphLink for other purposes (Copilot integration, data federation), this is not a barrier. For organizations evaluating GraphLink for the first time, it’s an additional initiative.
MDG Quality Must Be Sufficient
Kernaro AI Hub returns answers based on the quality of your EA governance. If your repository uses inconsistent tagging, informal element naming, incomplete relationships, or weak stereotype application, Kernaro AI Hub reflects that inconsistency. It doesn’t repair poor governance, it exposes it. Organizations with informally governed repositories need to either improve governance first or accept that Kernaro AI Hub’s answers will be imprecise. This is addressable (through Discover assessment and governance improvement), but it requires effort and timeline.
Additional Licensing Cost
Kernaro AI Hub licenses are purchased separately from Sparx Systems and carry their own cost structure based on user volume. For large organizations, this is an incremental platform cost in the $30K-$100K+ annual range. This is typically justified by interrupt cost savings, but it requires budgeting and CFO approval as a new initiative rather than an extension of existing EA spending.
Potential Parallel Investment with Copilot or Agentforce
If your organization is committed to Microsoft Copilot or Salesforce Agentforce for other reasons, Kernaro AI Hub may represent a parallel investment rather than a replacement. Copilot and Agentforce can both connect to EA repository data via MCP (Sparx Systems’ integration approach), serving stakeholders inside M365 or Salesforce tools respectively. Hub serves stakeholders with a dedicated architecture portal. Some organizations choose Hub for architecture depth, Copilot for M365 reach, or both. Others choose Copilot as the single integration point. This decision involves tool consolidation vs. specialized intelligence tradeoffs and may not justify Hub as an incremental investment.
Low-Volume Question Environments
If your stakeholder base asks fewer than 10-15 architecture questions per week, the interrupt cost is low enough that Kernaro AI Hub’s deployment investment and licensing cost may not be justified. You’re in the zone where one architect managing a question queue is sufficient. Hub is designed for organizations that have outgrown that model.
Before proceeding with Kernaro AI Hub deployment, confirm these prerequisites:
Enterprise Architecture Repository
EA GraphLink Deployed and Operational
MDG Technology Sufficiently Governed
Kernaro AI Hub License Purchased from Sparx Systems
Pro Cloud Server Configuration for Remote Access
The financial case starts with measuring architect interrupt cost:
Baseline Measurement:
Example calculation:
Secondary Measurement: Beyond capacity recovery, measure executive satisfaction with architecture visibility. After Kernaro AI Hub deployment, survey stakeholders on questions answered in real time vs. those that previously required architect requests. This is harder to quantify but often becomes apparent: executives who previously waited 2-3 days for architecture input now make decisions with live data. The impact on decision speed and quality is real but measured in decision cycle time, not hours saved.
Tertiary Measurement (Qualitative): Architecture team perception. Architects often report that the reduction in interrupt work increases job satisfaction and creates space for longer-term work they couldn’t reach before. This manifests as better roadmaps, more strategic analysis, and higher-quality recommendations. It’s not easily quantified but directly affects the practice.
Microsoft Copilot with MCP Integration Copilot can connect to the same EA GraphLink instance that Kernaro AI Hub uses. Stakeholders who live in Teams, Outlook, or Copilot Chat can ask architecture questions there and receive answers from live EA data. Pros: reaches M365 users in familiar tools; no new platform to learn. Cons: Copilot is general-purpose, not architecture-specialized; requires M365 licensing; integrates into M365 data management and security models rather than standalone architecture platform.
Salesforce Agentforce Agentforce can connect to EA GraphLink via MCP. Stakeholders in Salesforce organizations get architecture data inside Salesforce. Pros: reaches Salesforce users in familiar tools. Cons: requires Salesforce licensing; Agentforce is general-purpose; best suited for organizations where Salesforce is already the primary platform.
Prolaborate Prolaborate is a browser-based EA portal for browsing and searching the repository. It offers navigation and search but not natural language querying. Pros: lower cost; simpler to maintain; familiar interface for EA users. Cons: requires user training on EA structures and terminology; no natural language; questions that require aggregation or complex filtering are difficult; not suitable for stakeholders unfamiliar with EA.
Periodic PDF/Excel Reports Some organizations produce quarterly or monthly architecture reports (application portfolios, capability maps, technology roadmaps) as static documents and distribute them. Pros: zero ongoing cost; low technical requirements. Cons: reports are stale; ad-hoc questions still interrupt architects; stakeholders are left asking questions about outdated data; doesn’t scale to frequent question volume.
What questions can stakeholders ask Kernaro AI Hub?
Kernaro AI Hub can answer questions about applications, capabilities, infrastructure, dependencies, compliance status, standards coverage, technology inventory, risk factors, and relationships between elements in your EA model. It works best on questions that have a factual answer in the repository: “What applications support X capability?” “Which applications are end-of-life?” “What technologies are deprecated?” It works less well on questions requiring human judgment: “Should we retire this application?” “Is this architecture pattern good?” “What does the business want to do?” Hub is a repository query tool, not a strategy advisor.
Does Kernaro AI Hub require an EA license?
No. Stakeholders accessing Kernaro AI Hub do not need an Enterprise Architect license. Hub is a separate browser-based portal for non-architects to query the repository. Architects and modelers use EA directly; stakeholders use Kernaro AI Hub.
How does Kernaro AI Hub differ from Microsoft Copilot connected to EA?
Both can connect to the same EA GraphLink instance and serve the same repository data. The difference is interface and reach. Kernaro AI Hub is a dedicated stakeholder portal specialized for architecture questions. Copilot is a general-purpose assistant that integrates architecture data alongside email, documents, calendar, and other M365 content. Copilot reaches stakeholders inside Teams and Outlook (requires M365 licensing). Kernaro AI Hub reaches stakeholders outside M365 with a dedicated architecture interface. Some organizations use both: Hub for architecture specialists, Copilot for M365-centric users. The choice depends on your user distribution and tool preference.
How does MDG quality affect what Kernaro AI Hub can answer?
Kernaro AI Hub queries the EA repository schema as transformed by EA GraphLink using your MDG Technology definition. If your MDG is well-governed, stereotypes applied consistently, tagged values populated reliably, relationships properly maintained, Hub’s answers are precise and authoritative. If your MDG is informal or inconsistently applied, Hub’s answers reflect that inconsistency. For example, if you have an Application element tagged with “EndOfLifeDate” in 60% of cases but not 40%, Hub will return an incomplete end-of-life list. This is why Discover (MDG readiness assessment) is a prerequisite step.
Can Kernaro AI Hub connect to the repository without EA GraphLink?
No. Kernaro AI Hub requires EA GraphLink as the data transformation and connectivity layer. GraphLink transforms the physical EA repository schema using your MDG Technology definition, creating a queryable knowledge graph that Hub uses. GraphLink is not optional for Hub deployment.
What languages does Kernaro AI Hub support for queries?
Kernaro AI Hub accepts natural language queries in the language you specify (typically English, but check current product documentation with Sparx Systems). Answers are returned in the same language. Repository content (element names, descriptions, tagged values) should be in the query language for best results.
Is Kernaro AI Hub a Sparx Services product or a Sparx Systems product?
Kernaro AI Hub is developed and published by Sparx Systems. Sparx Services deploys it through the Connect service, provides governance guidance, and integrates it into your architecture practice. We are not the software vendor. Clients purchase Kernaro AI Hub licenses directly from Sparx Systems. Sparx Services provides the implementation service and ongoing practice development.
How long does it take to deploy Kernaro AI Hub through the Connect service?
Connect typically requires 12-16 weeks from agreement to production. Timeline depends on MDG governance readiness (assessed in Discover), security review and authentication integration requirements, and concurrent other initiatives. Organizations with strong governance and simple security requirements can complete faster; those with governance improvement requirements or complex security models require longer. We establish a specific timeline and milestone plan during the Connect discovery phase.
What happens when a stakeholder asks a question the repository can’t answer?
Kernaro AI Hub will either return “no results found” or attempt to interpret the question and return partial results. It doesn’t fabricate answers or guess. If a stakeholder asks “What applications support the X capability?” and the repository doesn’t have X defined as a capability, Hub returns no results. If the query is ambiguous, Hub may ask for clarification. The key is that Hub is only as good as the repository it queries. This reinforces the MDG readiness prerequisite, incomplete or inconsistent governance means incomplete or inconsistent answers.
Kernaro AI Hub is deployed through the Connect service. Connect includes MDG readiness assessment (or assumes prior Discover completion), EA GraphLink configuration, security integration, stakeholder portal configuration, and production launch.
If you’re uncertain whether your repository is ready for Kernaro AI Hub, start with Discover to assess MDG governance readiness and develop a roadmap to deployment.
Contact the Sparx Services team to discuss your Kernaro AI Hub deployment timeline and requirements.
Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.