03
Assessment

Defensibility Stack Assessment (A.G.E.N.T.)

A diagnostic that scores competitive durability across the five compounding layers of the A.G.E.N.T. stack — Architecture, Governance, Evolution, Network, Trust — and flags which spoke, if missing, will collapse the wheel under load.

ShareXLinkedInFacebookEmail
The five compounding moats
Architecture
Governance
Evolution
Network
Trust

What it does

Chapter 7 opens on two founders, same investor, same question — "What prevents a well-funded competitor from replicating your architecture?" — and opposite fates. Priya's Meridian Trade Finance had built "something that got harder to compete with as time passed." The other founder had not. The difference wasn't technology; both had the same foundation models. The difference was the A.G.E.N.T. Defensibility Stack: Architecture, Governance, Evolution, Network, Trust.

This tool turns that five-layer framework into a personalized diagnostic. It scores your organization on each layer, composites them into an overall defensibility score, plots a five-axis profile, and — most importantly — applies Chapter 7's hardest rule: the layers don't add, they multiply. As the chapter puts it, "This isn't a framework where you can score 4 out of 5 and succeed. A wheel missing a spoke doesn't spin slower. It fails entirely under load." Olive AI is the cautionary case: strong Architecture and Governance, but Trust never scaled past pilots, so Network never emerged, and there was nothing to defend when cash ran out.

The assessment exists because AI-Born moats are invisible. They have, in the chapter's words, "no line on a balance sheet." You can't point to factories or patents. You have to ask harder, fakeable-resistant questions: How fast does the system learn? How deeply is it embedded in customer workflows? How much regulatory judgment is crystallized into executable policy? This tool asks those questions and returns a defensible answer.

Who it's for: founders preparing for a raise, investors running diligence on a defensibility claim, and operators who need to know which layer to harden next before a well-capitalized competitor arrives.

Figure: The A.G.E.N.T. stack this tool operationalizes — five layers that multiply rather than add; miss one and the cycle breaks.

Each answer is normalized to a 0–100 sub-score using Chapter 7’s bands. The layers don’t add — they multiply. Updates live.

Load a Chapter-7 archetype
A · Architecture
59

Low Cognitive Overhead, zero-touch routine workflow. The foundation that sits beneath the rest.

Cognitive Overhead Index of your core workflow59
38 COI
Lower is better. Meridian rebuilt to 24; competitors sat at 71+. Score inverts COI — never double-invert.
% of routine workflow steps requiring zero human intervention65
65%
Direct: the share of the routine path that runs without a person in it.
Distinct human-intervention points in the core workflow60
4 points
Each intervention point is overhead. Fewer is stronger.
% of operating decisions expressed as code/policy, not tribal knowledge55
55%
Architecture that lives in people doesn't compound. Direct.
% of the core workflow that is fully reproducible from versioned config60
60%
If you can't rebuild it deterministically, a competitor can't be stopped from out-rebuilding it. Direct.
Time to stand up the core workflow from scratch in a new region/segment58
8 weeks
Lower is stronger — architecture that redeploys fast compounds across markets.
% of marginal volume served at near-zero marginal headcount60
60%
A moat without headcount means volume scales without people. Direct.
% of the workflow with end-to-end automated observability/telemetry58
58%
You cannot harden what you cannot see. Direct.

“A wheel missing a spoke doesn’t spin slower. It fails entirely under load.” Any layer below 40caps your tier at Eroding — regardless of the weighted average.

Eroding
56A.G.E.N.T. Defensibility · 0–100

The layers hold, but the cycle is exposed. Harden the weakest before a funded competitor probes it.

Five-axis profile · compounding ring ≥ 80 · spoke floor 40
A · Architecture · w0.2559
G · Governance · w0.2057
E · Evolution · w0.2553
N · Network · w0.1548
T · Trust · w0.1561

Advantages dissipate under competitive pressure — AI-enabled-tier durability. A defensible position is available, but the architecture has to compound rather than merely exist.

Prioritized plan · broken spoke first, then weakest layers
N · NetworkNext 90 days · now 48

Track your Embedding Coefficient stage for each major customer. A Stage One relationship is not a moat. Design integration depth deliberately toward utility status.

Weakest input: “% of new logos sourced via existing-customer referral / network effect” (35). Meridian reached utility status; a Stage One relationship is not a moat.

E · EvolutionBy 6 months · now 53

Measure your Iteration Half-Life now. If your IHL exceeds 30 days, you're not competing with AI-Born firms. The bottleneck is usually decision authority, not infrastructure — pre-authorize amendment categories.

Weakest input: “% of strategy changes gated through a Risk-Twin / simulation before merge” (35). Meridian shipped ~70 iterations/yr at sub-7-day IHL; StellarPay shipped 4.

G · GovernanceBy 12 months · now 57

Build compliance architecture before you need it. Regulatory goodwill accumulated early is the one asset competitors cannot acquire by writing a larger check. Codify each regulatory interaction as executable precedent.

Weakest input: “% of past regulatory interactions codified as reusable executable precedent” (45). Meridian reached regulatory approval in under 3 weeks post-rebuild.

Operationalizes the A.G.E.N.T. Defensibility Stack framework.
Further reading
From the books
  • Book 1, Chapter 7 — "Moats Without Headcount: The New Defensibility Stack" (the A.G.E.N.T. layers, Meridian/StellarPay, the wheel-and-spoke rule, the Invisible Moat).
The Dispatch — N°01

Essays from
the lineage break.

New essays, framework studies, excerpts and pre‑order news. Sent rarely. Never noise.