# What is enterprise architecture context (and why is RAG not enough)?

> Canonical: https://decisionpen.com/blog/what-is-enterprise-architecture-context  
> Published: 2026-10-25 · DecisionPen · Living EA series (D6a)  
> Related docs: /docs/why-not-copilot-or-rag, /docs/architecture-decision-workspace, /docs/enterprise-ask, /docs/it-knowledge-continuum

## Summary

Enterprise architecture context is the curated, metamodel-aware model of the IT estate — entities, relationships, and How/Why evidence — that humans and agents share. Document RAG retrieves passages; it does not own that shared landscape. DecisionPen holds this as an Architecture Decision Workspace (Context of Record).

## What do people mean by “context”?

Three layers are often conflated:

1. **Session context** — prompt window, chat history, tools for one run.
2. **Retrieval context** — passages from wikis/PDFs via similarity (RAG).
3. **Enterprise architecture context** — governed landscape model with evidence and accountable curation.

Layers 1–2 help assistants. Layer 3 is required for trustworthy architecture decisions and agentic delivery.

## Operating definition

**Enterprise architecture context** = metamodel entities and relationships + bound How/Why evidence + human curation + an askable interface. Without ownership and curation it is a retrieval pile, not a Context of Record.

## Why document RAG is not enough

RAG answers “which passages look similar?” Architecture must answer “what is authorised to be true of our landscape?” RAG alone does not maintain a metamodel, resolve conflicting sources, record rejections, or provide auditable provenance. See also: Copilot/RAG vs Architecture Decision Workspace; metamodel agent vs writing Copilot.

## Architecture context vs CMDB vs document stores

| Store | Strength | Gap for decisions |
| --- | --- | --- |
| Wiki / document KB | Narrative | No single landscape authority |
| CMDB / inventory | Operational assets | Weak How/Why and capability framing |
| Classic EA repo | Structured What | Often hollow without continuous evidence (Discovery Tax) |
| Architecture context | What + How/Why, curated, askable | Requires stewardship and a living method |

## How context is shaped

1. Discover — propose from documents, dialogue, inventory (automated EA discovery).
2. Structure — align to a metamodel (LeanIX-aligned where relevant).
3. Bind evidence — keep sources with entities.
4. Curate — humans accept, edit, reject, or defer.
5. Ask and publish — Enterprise Ask and generated artefacts consume the same Context of Record.

## Why this matters for agentic delivery

Build/infra/change agents need reliable landscape slices. Without a shared architecture context plane, every agent re-assembles the estate (Context Tax). DecisionPen holds architecture context; delivery agents consume it — DecisionPen is not a coding Copilot.

## How DecisionPen provides architecture context

Architecture Decision Workspace: metamodel-aware map; evidence-linked entities; human curation of AI proposals; Enterprise Ask with citations. Trial: https://decisionpen.com/trial

## When is DecisionPen the wrong tool?

General writing Copilot only; pure ops CMDB; freeform Visio/Lucid/BPMN as the primary job.

## FAQ

**What is enterprise architecture context?**  
Curated shared representation of apps, capabilities, tech, integrations, and decisions with How/Why evidence — the architecture Context of Record.

**Why is document RAG not enough?**  
Similarity over text ≠ owned metamodel, conflict resolution, or auditable landscape truth.

**How is it different from a CMDB?**  
CMDB lists assets; architecture context adds design relationships, rationale, and askable decision support.

**Who should own it?**  
EA (or equivalent steward) owns curation; teams and agents consume slices and contribute accepted updates.

**How does DecisionPen provide it?**  
ADW: structured entities + evidence + curation + cited Ask on the living map.

**When is DecisionPen wrong?**  
When the job is only summarisation, ops CMDB, or freeform drawing.

## Related posts

- https://decisionpen.com/blog/what-is-the-context-tax-in-enterprise-it  
- https://decisionpen.com/blog/how-ai-accelerates-automated-enterprise-discovery  
- https://decisionpen.com/blog/why-ea-must-own-enterprise-context-in-the-agentic-era  
- https://decisionpen.com/blog/what-is-automated-enterprise-architecture-discovery  
- https://decisionpen.com/blog/ai-enterprise-architecture-copilot-vs-metamodel-agent  
- https://decisionpen.com/blog/how-to-generate-architecture-docs-from-living-context  
- https://decisionpen.com/blog/why-ea-factsheets-miss-how-and-why
