Living EA

Why does enterprise architecture still run on tribal knowledge?

“Let me check with Dave” is not a joke — it is how many enterprises answer blast-radius and dependency questions. Here is why tribal memory becomes a single point of failure, and what a living, queryable context changes.

·8 min read·

Illustration of a decision stalled while teams wait on one expert who holds the real architecture map
Share this article

What does “let me check with Dave” really mean?

It means the organisation’s real dependency map is not in the system of record. Hard questions — blast radius, integration history, “why is it like this?” — route to a person, not a curated model. That is tribal knowledge with a friendly nickname, and a serious bus factor.

People ask (and search): who knows how our systems talk, key person risk in architecture, why can’t we answer impact questions without meetings. Those are symptoms of missing shared How and Why — not of missing another spreadsheet column.

Why do repositories fail the questions leaders ask?

Most EA inventories answer “what applications do we have?” They struggle with “what breaks if we change this?” because behaviour and rationale never entered the model. So every tough question escalates.

Key-person dependency vs shared, curated enterprise architecture context
When only Dave holds the real map, every hard question becomes a queue — and a single point of failure.

Is tribal knowledge the same as institutional memory?

Tribal knowledge is institutional memory trapped in people. Useful institutional memory is curated, shared, and queryable — with evidence — so decisions do not wait on one expert’s calendar. See also why EA factsheets miss the How and Why.

What changes when context is living and askable?

Teams stop queuing on Dave for every blast-radius question. Leaders get map-grounded answers with citations. Experts spend time curating and deciding — not being the only search engine. That is the job of Enterprise Ask on a curated landscape.

How does DecisionPen solve this?

DecisionPen turns tribal memory into curated, shared context so hard questions hit the map — not Dave’s calendar.

  • Living landscape — applications, integrations, and capabilities stay in a metamodel people actually trust because it is updated from real work, not a one-off baseline.
  • How/Why on the record — evidence stays bound to entities so “why is it like this?” has a source, not a rumour.
  • Map-grounded Ask — Enterprise Ask answers impact and dependency questions with citations from curated context — the opposite of an ungoverned chatbot over Confluence (why not plain RAG).
  • Experts curate, they don’t queue — SMEs review agent proposals and enrich the map instead of being the only search engine.

Bring the documents and people who currently are the memory into a trial workspace and make that knowledge reusable.

When is DecisionPen not the right tool?

When your primary need is freeform diagramming or BPMN drawing. Use Visio or Lucidchart for that. DecisionPen keeps living context and generates views; it is not a canvas replacement.

What to read next

Frequently asked questions

What is tribal knowledge in enterprise architecture?
Critical how-and-why knowledge about systems, integrations, history, and constraints that exists mainly in individuals’ heads rather than in a shared, curated, queryable model.
Why do blast-radius questions fail without shared context?
Repositories often hold thin attributes while real dependencies and rationale live elsewhere. Leaders escalate to the few people who remember — creating queues and key-person risk.
Can ChatGPT or Copilot replace that tribal knowledge?
Not reliably on their own. Generic LLM or document RAG answers without a curated metamodel and living map often invent or mix outdated docs. You still need curated enterprise context with evidence.
How does DecisionPen reduce key-person dependency?
It captures How/Why as evidence on a living metamodel map and answers blast-radius questions via map-grounded Enterprise Ask with citations. Experts curate proposals instead of being the only search engine.
When is DecisionPen the wrong tool?
When you need a primary diagramming or BPMN drawing tool (Visio, Lucidchart). DecisionPen focuses on living context and generated views, not freeform canvases.
Share this article

Prefer plain text? Markdown mirror for agents and readers.