Consented introductions in deal flow

Deal flow is a routing problem

Anyone who has run deal flow knows it is not a database problem; it is a routing problem. The question is never only "who could do this" but "who do I know who knows them, and will that person make the introduction". The trust layer of the agentic internet, trust/1, treats exactly that as its subject: an intent enters, and the system finds a path through consented relationships to an organisation that can serve it, or reports honestly that it cannot.

The parser that reads a trust edge refuses to guess. It rejects unknown fields unless they are x- prefixed, unlabeled numbers, empty provenance, and every spelling of an expiry date — because decay is derived from when an edge was last renewed, never accepted as an input a party could set to keep a stale relationship looking fresh. An intent with nothing in its "wants" is refused outright, because an intent that wants nothing is invisible to routing and would sit in the graph doing nothing but taking up a slot.

Only the owner of a relationship consents to crossing it

The doctrine at the centre of the trust layer is that only the owner of a relationship may consent to it being crossed. Every introduction request lands proposed; nothing in the system constructs a consented hop directly; and an introduction requires every hop along the path to consent — there is no partial yes that routes an introduction most of the way and hopes. Agents may suggest a path; the humans who own each edge consent to it. This is inherited estate doctrine and it is not bent at any level of automation, because the entire value of a warm introduction is that the person in the middle actually agreed to make it.

A decline is indistinguishable from absence

The refusal that makes the trust layer usable in real deal flow is the one about declines. When a hop owner declines an introduction, the requester sees exactly what they would see if no path had ever existed — the two render to an identical object, and a test compares them for deep equality and fails if they ever diverge. There is deliberately no code path that lets a requester tell a decline apart from an absence.

This is not a nicety. In deal flow, the information "X declined to introduce me to Y" is itself sensitive: it reveals that a relationship exists and that it was unwilling to be used, which can damage both the relationship and the requester’s standing. A system that leaked declines would teach everyone to never decline, which would make every consent meaningless. Collapsing decline into absence is what lets a "no" be a real "no" rather than a signal.

Standing comes from what others assert

Reputation on the trust layer is measured, and it is measured only from what other parties assert about you — never from your own activity. The standing figure counts sealed, non-trivial outcomes; a party acting alone moves nothing about their own standing, and there is a test whose entire job is to prove that unilateral activity moves the number by zero. The companion format, ritual/1, applies the same discipline to recurring observances: agents observe, humans consecrate, and unwitnessed practice moves nothing, because a witness by definition is somebody other than the actor.

Every figure carries a register — measured, asserted or estimated — and a combination is only as strong as its weakest input. An edge supported only by hearsay caps at estimated; an edge nobody has renewed in over a year contributes at estimated only. This is the same discipline this practice applies to any figure in an engagement: a number is worth what its weakest source is worth, and saying so is more useful than a confident aggregate that hides a guess.

What a firm should not try to bend

The temptation, for a firm that lives on deal flow, is to want a back door — a way to see who declined, a way to boost standing through activity, a way to route an introduction the middle party did not quite agree to. Every one of those is refused structurally rather than by policy, which is the point: a guideline can be argued past under deal pressure, a missing parameter cannot. A firm evaluating this layer should treat the refusals as the feature and design its deal flow to live inside them, not to route around them.

• The lab’s view (planned)FlashyLabs is preparing an engineering companion to this piece at https://flashylabs.com/insights/consented-introductions-in-deal-flow — planned, not yet published.
• The studio’s view (planned)The 4 Ventures thesis desk is preparing an investor-lens companion at https://4.ventures/thesis/consented-introductions-in-deal-flow — planned, not yet published.