# Make Them Insurable: The Positioning, And The Survey That Would Test It

*Doctrine document 09. Derived from memo 9 ([brief v0.33.80](../briefs/v0.33.80__strategy-brief__make-your-agents-insurable-and-the-first-fact-this-pivot-would-produce.md)). Proposed, not adopted: GM-D72 to GM-D75 await the project lead.*

---

## 1 · The first thing here that could be wrong

**As of memo 9 this pivot had produced seventy-one decisions and no external evidence**, and the count has only grown since. Every document in this folder is reasoning — checked against the corpus, and entirely internal.

*Can you insure a Claude Code session?* is different in kind: **the answer exists outside this estate and nobody here has looked.**

> **The survey is the first thing in this series that could be wrong in a way the world would correct.**

For a body of work whose own rule is that a claim must be checkable, that matters more than any position in the preceding nine documents.

## 2 · The positioning, and its guard-rail

**"Make your agents insurable"** places the work as *preparation for a market* rather than a product in one — honest, since the market does not exist, and useful, since it does not require the market to exist before the work has value.

The guard-rail mirrors this folder's rule about levels:

> **"Make your agents insurable" must not become "make your agents look insurable."**

Insurability is achieved by **narrowing the delta or covering it** — never by documenting it. A programme that helped every placement become insurable would be doing exactly what a rating that only ever improves does.

And the honest counterpart, said out loud: **some placements should not be insurable.** A desktop agent under a user identity with network egress and no containment may be uninsurable at any level, and **the useful output is that sentence.** Finding one is a success of the method, not a failure of the customer (GM-D72).

## 3 · What breaks cover — and the mapping that upgrades a rule

The inverse question defines cover more sharply than the cover does, and insurance has three answers. Two the estate already carries; **the third is new.**

| What breaks cover | In insurance | Here |
|---|---|---|
| **Exclusions** | Never covered, stated up front | The mandate's **prohibitions** |
| **Warranties / conditions** | True at inception and maintained; breach voids | The **facts** — a breached warranty voids cover exactly as a flipped fact drops the enforcement tier |
| **Material non-disclosure** | You did not disclose something that would have changed the terms | **The gap between the declared channel and the measured channel** |

**The third is the finding.** [Document 01's](the-rating.html) two channels have been a *rating accuracy* rule. They are also a *voidability* rule:

> **A control declared on the card that the twin cannot find is not merely a weak input. It is the shape of material non-disclosure** — the thing that voids a policy rather than merely worsening a level (GM-D73).

Which changes what the card-versus-twin gap is worth computing for. It has been an input; it is also **the single most consequential quantity an underwriter would want**, because it decides whether a claim gets paid. And the estate can compute it today: the twin and the card are both documents it already holds.

## 4 · The survey is a measurement, not a report

*As the market evolves, you can keep track of it* — so this is a **dated, re-runnable observation**, and the estate knows how to build one honestly.

> **The survey is `measure.py` pointed at a market instead of a container**, and inherits its discipline unchanged (GM-D74):

| The measurement rule | Applied to a market |
|---|---|
| Presence and reachability, never contents | Record what a product **states publicly**; do not characterise what it privately offers |
| Every node carries its **evidence class** | `observed` (a published price), `read` (a policy wording), `documented` (a vendor claim), `inferred`, `none` |
| **Unknown is never absent** | *"We could not establish whether X covers agent placements"* is the finding — not *"X does not"* |
| A grant is a **floor, not a census** | The survey is a floor: what was not found may still exist |
| The measurement is dated, and its age printed | An insurance observation goes stale fast; a survey with no date is a claim about now that was true once |

**And the same self-restraint the tool has.** `measure.py` reports presence and never contents so it cannot leak what it looked at. The market analogue: **report what vendors publish, not what somebody said in a sales call** — the second is unverifiable by a reader and makes the survey unrepeatable by anyone else.

## 5 · What this document will not guess

Four questions in the memo have real answers, and this document does not know them.

The **categories** that exist generally can be named — technology errors-and-omissions, entity-level cyber cover, vendor indemnities commonest for intellectual-property claims — and **none of them insures an agent placement**, which is the gap the whole pivot assumes. But whether some carrier launched an agent product last month is exactly the kind of fact that would be invented rather than found.

> **Predicting the survey's answers and publishing them as research would be this estate's cardinal sin**, committed in the one document whose entire value is that somebody went and looked.

The prediction, stated once so it can be falsified: **the survey will find no product insuring an agent placement as a unit.** If that holds, it is the strongest argument for the positioning. If it does not, the positioning changes — which is what evidence is for.

## 6 · Evidence is where this estate is placed, and where the survey pays

Every process the memo lists — web form, spreadsheet, Word document, API — collects evidence **by questionnaire**: the declared channel. This estate produces **measured** evidence with provenance, which none of them asks for.

That is half an advantage. The other half:

> **An underwriter does not want a JSON document they cannot interpret.** The gap is not producing evidence; it is producing evidence in a form an existing process can consume.

So the survey's most commercially valuable output is not prices or coverage terms but **the format the market accepts** — because that format is the connector specification (GM-D75). [Document 05](not-in-line.html) already requires a connector to label the evidence class of what it imports; this is the same requirement pointing outward, and it needs the survey to know what it connects *to*.

**And the first real deliverable follows:** an evidence pack, plus a rendering into whatever the market fills in today — where the rendering **must carry the evidence classes across**, or a measured fact and a declared one arrive at the underwriter looking identical, which is §3's non-disclosure created by our own connector.

## What this does not prove

- **That any of the survey's answers are known.** §5 is explicit: the categories are background, the market's current state is unresearched, and the prediction is published as falsifiable rather than as a finding.
- **That the positioning is right.** *Make your agents insurable* presumes insurability is the goal; §2's guard-rail exists because that presumption is the thing most likely to bend the work.
- **That the non-disclosure mapping would hold legally.** §3 argues the card-versus-twin gap has the *shape* of material non-disclosure. Whether any carrier or court would treat it that way is unknown and is not this estate's competence.
- **That a re-runnable survey is affordable.** §4 specifies the discipline; keeping a market observation current is ongoing work nobody has costed.

---

*CC BY 4.0. Source: brief v0.33.80, memo 9. Everything here is derived from that memo and labelled where it extends it.*