---
description: >-
  The digital twin: a living, structured model of an engagement — goals, current state, decisions with reasons, and tracked unknowns — replacing tribal knowledge.
---

# The digital twin

When you arrive at a hospital, nobody asks whichever clinician happens
to be on shift to remember your history. There is a chart: allergies,
medications, what was tried, what was ruled out, which results are still
pending. The chart is why your care can change hands — between shifts,
departments, even hospitals — without you starting over from the
beginning, and without anything important depending on one person's
memory.

Enterprise delivery mostly runs without a chart. Every engagement
accumulates knowledge: what the customer wants and why, what already
exists in their systems, what was decided and for what reasons, what was
considered and rejected, and what nobody knows yet. Today that knowledge
scatters — a goal stated in a kickoff deck, a constraint buried in a
note file, a decision made in a chat thread, a caveat living in one
architect's head. When the engagement changes hands, whoever picks it up
reconstructs the history from fragments. When people leave, the parts
only they knew leave with them.

The **digital twin** is that knowledge kept as the chart instead: a
living, structured model of the engagement. Not a document that
summarizes the engagement at one moment, but a model that *is* the
engagement's current state of knowledge — updated as the work proceeds,
readable by anyone who joins, and never dependent on who happens to be
in the room.

## What the twin holds

In plain terms, four kinds of things:

- **The goals.** What the business has actually said it wants, and why —
  not a paraphrase of it three retellings later.
- **The current state.** What exists in the customer's environment today:
  the systems, the data, the constraints the work must respect.
- **The decisions.** What was decided, with the reasons attached — and
  what was ruled out, so the same dead end is not explored twice.
- **The open questions.** What remains unknown, tracked explicitly as
  part of the model rather than silently forgotten. An unanswered
  question in the twin is a visible item with a name, not a gap someone
  discovers in week nine.

Every entry also carries who added it and on what evidence — the
provenance discipline described in [Governance as trust](governance.md).

![The digital twin at center holding goals, current state, decisions, and open questions, inside a dashed provenance boundary](../assets/twin-model-light.svg#only-light)
![The digital twin at center holding goals, current state, decisions, and open questions, inside a dashed provenance boundary](../assets/twin-model-dark.svg#only-dark)

## Why it matters

The twin is not a filing system; it is the compiler's input. In
[governed intent compilation](governed-intent-compilation.md), the
[Spec Manifest](spec-manifest.md) — the governed plan that travels from
executive conversation to production configuration — is compiled *from*
the twin. The plan can only be as good as the model it is compiled from,
and that is precisely the point: because open questions are first-class
entries rather than omissions, a thin twin produces an honest, thin plan
that says what is still unknown — not a confident hallucination that
papers over the gaps. The quality of the model and the quality of the
plan are visibly, checkably connected.

That is the difference between tribal knowledge and an asset. Tribal
knowledge walks out the door; a twin compounds — every conversation,
decision, and discovery makes the model, and therefore every plan
compiled from it, better.

---

*Terms used across these pages are collected in the
[plain-English glossary](glossary.md).*