# How do you enrich EA factsheets with unstructured documents?

> Canonical: https://decisionpen.com/blog/how-to-enrich-ea-factsheets-with-unstructured-documents  
> Published: 2026-10-22  
> Author: DecisionPen Team  
> Category: Automated Landscape Modelling

## Short answer

Start with the structured skeleton you already have — typically an exported CMDB or application inventory — then attach HLDs, ADRs, runbooks, transcripts and expert knowledge to each entity. Use AI to propose purpose, integrations, rationale, constraints and risks, keep every claim traceable to evidence, and let architects curate the enrichment before it becomes part of the living model.

## Your CMDB knows what exists. It rarely knows why it matters.

A CMDB-derived factsheet can establish name, owner, status, environment, technology and criticality. It rarely explains business purpose, design rationale, integration behaviour, undocumented dependencies, constraints or change risk. Those answers generally live in documents, diagrams and people.

## What an enriched factsheet contains

1. **What:** structured identity — name, type, owner, lifecycle, environment, technology and criticality.
2. **Why:** business purpose, supported capabilities, decision drivers and rationale.
3. **How:** integrations, information flows, protocols, operational behaviour, constraints and dependencies.
4. **So What:** blast radius, risk, change readiness, technical debt, retirement barriers and recommended decisions.

## Worked example

A CMDB record for Customer Payments Hub might contain only its name, Payments Operations owner, Active status, Tier 1 criticality and Java/Oracle/IBM MQ technology. After enrichment from an HLD, ADR, runbook and workshop, the same entity can explain that it ingests, validates, orchestrates and settles payments; supports related capabilities; exposes APIs; integrates with risk and fraud over MQ; posts to core banking; and presents a data-consistency risk across ledgers.

## Practical workflow

### 1. Seed the structured skeleton

Export a bounded CMDB or inventory extract. In DecisionPen, import applications, technology components and similar catalogue items from a spreadsheet with guided column mapping. Normalise names and remove obvious duplicates.

### 2. Attach evidence to existing entities

Add HLDs, ADRs, interface specifications, runbooks, migration documents and transcripts. Match material to the imported entity before creating new entities.

### 3. Propose How and Why

Extract candidate purpose, capability links, integrations, data flows, decisions, constraints, risks and missing relationships. Keep source provenance visible.

### 4. Make conflicts visible

Do not let an algorithm silently choose between the CMDB and a newer document. Show source and date, distinguish observed facts from documented intent and inferences, and route conflicts to accountable curators.

### 5. Curate, then ask

Accept, edit, reject or defer proposed context. Trusted factsheets can then support questions such as: “What depends on this hub?”, “Why was MQ selected?”, and “Which capabilities would a ledger replacement affect?”

### 6. Keep enriching incrementally

New designs, workshops, programme decisions and operational knowledge should continue to enrich the model.

## Product boundary and CMDB accuracy

DecisionPen can start from a CMDB or inventory spreadsheet extract. It is not an operational CMDB replacement. It is also not a freeform Visio or BPMN canvas.

## How DecisionPen solves this

DecisionPen imports a structured catalogue through spreadsheet import, binds documents, transcripts and expert narratives to entities, proposes metamodel-aware enrichment and relationships, and supports human curation. The resulting Architecture Decision Workspace combines structured identity with evidence so Enterprise Ask can answer from both.

## FAQ

### Why are CMDB-derived factsheets incomplete?

They are strong on identity and operational attributes but weak on purpose, rationale, behaviour, undocumented dependencies and change implications.

### Can I start DecisionPen from a CMDB?

Yes, by exporting the CMDB or inventory catalogue to a spreadsheet and importing catalogue items.

### What happens when documents contradict the CMDB?

Preserve each claim with source and date, flag the conflict, and ask the accountable architect or owner to curate the accepted position.

### How does DecisionPen enrich factsheets?

It matches evidence to imported entities, proposes How-and-Why context and relationships, retains provenance and uses human curation before information joins the living model.

## Related reading

- https://decisionpen.com/blog/why-ea-factsheets-miss-how-and-why
- https://decisionpen.com/blog/automated-it-landscape-modelling-from-unstructured-data
- https://decisionpen.com/blog/it-landscape-discovery-guide
- https://decisionpen.com/docs/incremental-modelling
- https://decisionpen.com/use-cases/it-landscape-mapper