§ 01 ── HOW IT WORKS ── ARCHITECTURE

Eleven agents. One audit chain.
A methodology overlay that travels with your fund.

Six customer-operational specialists. Five architectural agents that keep the substrate substrate-grade. One overlay that's yours. The architecture is segment-agnostic. For deeper architectural doctrine, read /substrate →

§ 01.1 ── ARCHITECTURE ── TWO LAYERS

System of systems.

L · FUND LAYER11 AGENTS · OPERATED BY THE FUNDoverlay · audit chain0102030405AUDIT CORE0607080910— 6 SPECIALIST5 ARCHITECTURAL —COMPOSE · CONSENT-GATED · k ≥ 8L · PORTCO LAYER · OPTIONAL11 AGENTS · OPERATED INSIDE EACH PORTCOfounder consent · revocable in 24h0102030405AUDIT CORE0607080910— 6 SPECIALIST5 ARCHITECTURAL —

FIG. 01 ── FUND LAYER · CONSENT-GATED COMPOSE · PORTCO LAYER

LEFT · FUND LAYER

Eleven agents at fund tier. Six customer-operational specialists handle sourcing, diligence, IC memos, founder support, LP reporting, portfolio strategy. Five architectural agents — Audit Curator, L8-Enforcer, Registry Maintainer, Eval Suite Runner, Version Controller — keep the substrate substrate-grade. All operate within a methodology overlay encoding your specific scoring frameworks, sub-thesis weights, LP voice tiers, regulatory posture.

RIGHT · PORTCO LAYER

Optional. When a founder consents, a per-company instance runs inside their tenant — eleven agents in parallel architecture: six domain operational (Finance, Marketing, Product, Engineering, People [privacy-bounded], Strategy) + five architectural. Aggregate signals flow back to the fund layer through consent gateways at k ≥ 8 anonymity. Revocation halts data flow within 24 hours.

§ 01.2 ── TIER FRAMEWORK ── 4 LEVELS

Four tiers. Every agent occupies one.

Every agent in the substrate has a declared tier ceiling. For L8-floor agents, the higher-tier code path does not exist. Tier ceilings are architectural, not policy.

AUTHORITY ↑T1 · OBSERVERread-only · reportsT2 · DRAFTERdrafts · advisoryT3 · OPERATORdrafts + executesT4 · STEWARDsubstrate authority

FIG. 02 ── OBSERVER · DRAFTER · OPERATOR · STEWARD

TIER
NAME
AUTHORITY
EXAMPLES
T1
OBSERVER
Read-only · reports findings · cannot draft outputs.
Pipeline scanners · sentiment monitors.
T2
DRAFTER
Drafts outputs for human review · cannot commit · advisory only.
Investment Principal (L8-cap) · Founder Support (L8-cap).
T3
OPERATOR
Drafts AND executes defined operations within bounded scope · still human-reviewed.
Sourcing · Diligence · LP Reporter · Portfolio Strategist.
T4
STEWARD
Holds authority over substrate discipline itself · L8 enforcement · audit chain integrity · versioning.
L8-Enforcer · Audit Curator · Version Controller · Eval Suite Runner.
§ 01.3 ── AGENTS · CUSTOMER-OPERATIONAL

Six specialists. Two with L8 hard caps.

OUTER RING · CUSTOMER-OPERATIONAL · 6INNER RING · ARCHITECTURAL STEWARDS · 501 · SOURCINGtier T302 · DILIGENCEtier T3L8 CAP03 · INVEST. PRINCIPALtier T2L8 CAP04 · FOUNDER SUPPORTtier T205 · LP REPORTERtier T306 · PORTFOLIO STRAT.tier T301 · AUDIT CURATORtier T402 · L8-ENFORCERtier T403 · REGISTRYtier T404 · EVAL SUITEtier T405 · VERSION CTRLtier T4AUDIT CURATORsole-emission

FIG. 03 ── 6 SPECIALISTS · 5 STEWARDS · ONE SOLE-EMITTING CORE

01
Sourcing Analyst
TIER-3 (Operator) ceiling
Inbound triage · cohort matching · evidence validation · the first-line filter that multiplies partner throughput.
02
Diligence & Impact Lead
TIER-3 (Operator) ceiling
Hexframe-class scoring · counter-evidence search · sub-thesis routing · evidence-base integration (PubMed · ClinicalTrials · NICE · Cochrane).
03
Investment Principal
TIER-2 (Drafter) · L8 HARD CAP
IC memo synthesis · objection mapping · cap-table modeling. Never autonomous on investment recommendations. Human-in-loop enforced at the architecture level.
04
Founder Support
TIER-2 (Drafter) · L8 HARD CAP
Wellbeing extraction (consent-bounded) · Care-to-Dare candidacy · Soft Landing triggers. Never reaches founders directly — surfaces to partners only.
05
LP Reporter
TIER-3 (Operator) ceiling
Per-tier voice calibration · Article 9 SFDR audit chain on every claim · per-LP-tier framing.
06
Portfolio Strategist
TIER-3 (Operator) ceiling
Cross-portfolio matcher · prior-decision retrieval · cohort-context retrieval · federation activates at k ≥ 8 portcos.
§ 01.4 ── AGENTS · ARCHITECTURAL

Five Tier-4 Stewards. Most engagements never think about them. That's the point.

The architectural agents keep the substrate substrate-grade. They surveil L8 floors, enforce audit chain integrity, manage versioning, and run κ verification. Most customer engagements never directly interact with them — they operate as the substrate's immune system. But without them, "substrate-grade" is a claim, not a property.

01
Audit Curator
Tier-4 Steward
Sole-emission authority per tier · chain integrity by construction · severity classification logic · NEVER-deletable enforcement at filesystem level.
02
L8-Enforcer
Tier-4 Steward · recursively self-protected
Continuous surveillance of all L8 floor categories · refusal events emit severity:critical · quarterly self-tests (5/5 PASS expected) · cannot be silenced by other agents.
03
Agent Registry Maintainer
Tier-3 Operator
Weekly sweep · schema validation per agent contract · 4-surface alignment · cross-cluster reference resolution · orphan detection.
04
Eval Suite Runner
Tier-4 Steward
Operates §10 canonical eval cases · κ ≥ 0.85 verification on scorer agents · halt authority on κ drift detection.
05
Agent Version Controller
Tier-4 Steward
MAJOR / MINOR / PATCH cycle enforcement · pre-snapshots · replay-tests · stakeholder concurrence routing · L8 floor amendment requires BOTH stakeholder concurrence AND domain-architectural redesign evidence.

The substrate's immune system. Invisible when working. Load-bearing when stressed.

§ 01.5 ── ARCHITECTURE ── AUDIT CHAIN

Append-only. Hash-chained. Sole-emission. Replayable. Governance-grade.

Every system action — every score, every memo, every LP draft, every wellbeing inference — writes to an append-only audit chain via the Audit Curator (sole-emission authority per tier). Each entry hash-links to the prior entry. Tampering breaks continuity end-to-end and is detectable. Beyond defensibility (explaining past decisions), the audit chain enables governance (constraining future ones).

AUDIT CURATORsole-emission · per tierAPPEND-ONLY · HASH-CHAINED · NEVER-DELETABLETIME →T0042infoprev 7c4f…self 3b8d…overlay v1.4.2T0043infoprev 3b8d…self 9a12…overlay v1.4.2T0044infoprev 9a12…self f02e…overlay v1.4.2T0045infoprev f02e…self 1c66…overlay v1.4.2T0046infoprev 1c66…self 8e44…overlay v1.4.2

FIG. 04 ── ONE CURATOR · FIVE BLOCKS · CONTINUITY BY CONSTRUCTION

01

SOLE-EMISSION

Exactly one Audit Curator agent per tier holds emission authority. All other agents route emission requests through the curator. Chain integrity by construction, not policy.

02

REPLAY

Given an audit ID, the substrate reconstructs the exact context that produced the original output — overlay version, model version, evidence set — and reproduces it bit-for-bit.

03

SCHEMA MIGRATION

Overlay version-tagged on every entry. Migrations replay against last 100 entries before promotion. Past decisions stay defensible across overlay updates.

04

NEVER-DELETABLE

Severity:critical entries are filesystem-level append-only. No delete API exposed. Deletion attempts emit a new severity:critical entry (recursively) and themselves persist forever.

05

ARTICLE 9 SFDR

Every impact KPI movement timestamped, evidenced, traceable. LP letters cite audit chain IDs inline. Regulator-defensible by architecture, not by post-hoc reconstruction.

// Example · audit chain entry (emitted by Audit Curator)
2026-Q3-W4-T0042 | agent=diligence_lead | overlay_v=1.4.2 | model=opus-4.7
  input_hash:        f8a3c7b9
  output_hash:       2e4d1a06
  evidence_set:      [pubmed:34521098, clinicaltrials:NCT04812345, nice:NG185]
  hexframe_composite: 8.17
  partner_overrides: [{sub_dim: P_2, delta: -0.4, rationale: "founder_signal_weak"}]
  txn_id:            TXN-2026-Q3-W4-0017
  severity:          info
  hash_prev:         7c4f2a8e
  hash_self:         3b8d6c5f
  emission_curator:  audit_curator_fund_tier_v1

Every field is queryable. Every entry is replayable. Every claim carries lineage.

§ 01.6 ── L8 FLOORS ── CODE-PATH ABSENCE

Commitments encoded as code-path absence.

L8 architectural floors are commitments that cannot be lifted by policy. They are encoded as the absence of the lifting code in the substrate — not as runtime checks, not as authorization roles, not as configuration flags. The L8-Enforcer surveils continuously. Refusal events emit severity:critical NEVER-deletable. Self-tests run quarterly with 5/5 PASS expected.

── HIGHER-TIER CODE PATH (DOES NOT EXIST)── SUBSTRATE FLOORF1INVESTMENTPRINCIPALtier-2 capF2FOUNDERSUPPORTtier-2 capF3WELLBEINGPATHWAYdefault OFFF4FEDERATIONk-FLOORk ≥ 8F5CONSENTGATEWAY 2fail-closed

FIG. 05 ── FIVE PILLARS · NOTCHED CAPS · ABSENCE BY DESIGN

F1 · INVESTMENT PRINCIPAL AUTONOMY
Tier-2 (Drafter) hard cap. No Tier-3+ code path exists for autonomous investment recommendations.
F2 · FOUNDER SUPPORT TIER CEILING
Tier-2 (Drafter) hard cap. Output never reaches founders directly · partner-only surface.
F3 · WELLBEING PATHWAY DEFAULT-OFF
Founder consent default OFF · opt-in required · revocation halts within 24h · aggregates purge within 72h.
F4 · FEDERATION K-FLOOR
Cross-portfolio federation refused below k=8. Current k value displayed in operator state.
F5 · CONSENT GATEWAY 2 FAIL-CLOSED
Fail-closed default · per-category opt-in · revocation events audit-emitted at filesystem level.

Floors evolve only via MAJOR amendment cycle + domain-architectural redesign evidence + stakeholder concurrence. BOTH required. Either alone is refused.

§ 01.7 ── PROGRESSION

Observe → Explain → Constrain → Govern.

01OBSERVElogs · events02EXPLAINreplay · lineage03CONSTRAINthresholds · caps04GOVERNL8 floors · topology

FIG. 06 ── FOUR STAGES · ASCENDING PLATEAUS

01

OBSERVE

Logs, metrics, events.

Where most AI bolt-ons stop.

02

EXPLAIN

Replay, evidence lineage, audit chain.

Where mature audit-aware AI operates.

03

CONSTRAIN

Refusal under threshold · tier caps · escalation.

Where the substrate constrains its own outputs.

04

GOVERN

L8 architectural floors · authorization topology · schema lock · reversibility.

Where the substrate operates by design.

Defensibility explains past decisions. Governance shapes future ones.

§ 01.8 ── ARCHITECTURE ── CONSENT

Three layers. Each one revocable.

revoke01 · FUND-INTERNALdeal flow · IC · LP commsrevoke02 · FOUNDER-LEVELper-tenant · revocable in 24hrevoke03 · CROSS-PORTFOLIO · FEDERATIONk ≥ 8 anonymity (L8 floor F4)DATA ORIGIN

FIG. 07 ── CONCENTRIC BOUNDARIES · OUTWARD REVOCATION

01
FUND-INTERNAL
Deal flow · IC memos · LP communications. Operates under standard fund confidentiality. No data crosses this boundary except as the overlay defines.
02
FOUNDER-LEVEL
Each portco's data stays in its tenant. The founder consents explicitly to which aggregate signals flow back to the fund layer. Wellbeing extraction is a separate opt-in. Consent is revocable; flows halt within 24 hours; aggregates purge within 72 hours.
03
CROSS-PORTFOLIO FEDERATION
Patterns surface across portcos only above k ≥ 8 anonymity (L8 floor F4). With <8 portcos on the substrate, federation is inactive. At 8+, anonymized peer comparisons activate. Founders opt out individually; aggregates re-compute within 30 days.
§ 01.9 ── OVERLAY

Your methodology, in a single YAML file.

The Generic Architecture is fund-agnostic. The customer overlay is where your specific methodology lives: scoring axes, sub-dimensions, decision thresholds, LP voice samples, sub-thesis priorities, IAC composition, regulatory cadence. Every agent loads it at bootstrap. Every audit chain entry tags the version.

// customers/[fund].yaml · methodology overlay
overlay_version: 1.0.0  # MAJOR.MINOR.PATCH

scoring:
  framework: hexframe_v1
  axes: [AXIS_1, AXIS_2, AXIS_3]
  sub_dimensions: 6
  thresholds:
    thesis_fit:       7.0
    diligence_pursue: 7.5
    ic_recommend:     8.0

evidence_base:
  sources: [pubmed, clinicaltrials, cochrane, nice]
  counter_evidence_band: [0.85, 0.90]

kappa_verification:
  floor:   0.85   # Cohen's κ weighted variant
  cadence: weekly_primary_monthly_voice_quarterly_comprehensive

lp_voice_tiers:
  - { name: tier_1, samples: 10, framing: rigor_first }
  - { name: tier_2, samples:  8, framing: regulatory_cascade }
  - { name: tier_3, samples:  6, framing: returns_balanced }
  - { name: tier_4, samples:  4, framing: institutional }

governance:
  l8_floors:
    investment_principal: tier_2_hard_cap
    founder_support:      tier_2_hard_cap
    wellbeing_pathway:    default_off_opt_in_required
    federation_k_floor:   8
    consent_gateway_2:    fail_closed_default
  confidence_escalation_floor: 0.70

Axes are abstract placeholders until Phase 0 workshops resolve them. The overlay is versioned in git. You own it (L2 IP).

§ 01.10 ── BOUNDARIES

What we don't build.

SAAS
We don't sell a multi-tenant SaaS product. Each fund gets a dedicated instance in their own tenant. Your data is yours. Your overlay is yours.
COPILOTS
We don't build copilots. Copilots assist partners. We build substrate that operates as partners — with L8 architectural caps where it shouldn't.
GENERIC AI
We don't do generic AI consulting. No "AI for VC" templates. Each engagement starts with your methodology, not ours.
GENERALISTS
We don't work with generalist funds. The architecture compounds when methodology is segment-specific. Generalists don't have that compound advantage.

§ END ── DEEPER

If the architecture interests you, the doctrine is at /substrate.