§ 00 ── SCALING COMPANY OS · SERIES B / C
Capture functional methodology
before dilution.
For Series B/C companies. 30–200 employees. $3–30M ARR. Functional leadership is emerging. VPs carry methodology the founders never had bandwidth to develop. The substrate captures it — across leadership, across functions, across the cross-functional handoffs that are FRAGMENTING right now. Where you used to need middle managers routing information up and down the org chart, canonical Class A agents (T2 drafters) route. VPs judge at the edge — per Jack Dorsey's framing at Block: humans guiding, not middleware.
§ 01 ── YOU KNOW THIS REALITY
50–150 employees. VPs in place. Functional silos surfacing.
Your VP Eng has decision heuristics that differ from your founders'. Your VP Sales has qualification patterns the founders didn't have time to develop. Each functional leader carries methodology worth preserving. Without explicit capture, that methodology lives in their head and walks out when they leave. By Series C, “how we do things” is fragmented across teams.
At Seed-A, functional integration is the founder. At Series B/C, the founder is too constrained to be the integrator. The substrate becomes the integrator the founder used to be — and Diana Hu's YC framing applies: the org chart's information-routing layer migrates from middle managers into the intelligence layer.
The canonical 7-seat Methodology Council (formed at Phase 0 Week 2, standard at B/C scale — typically founders × 2 + 3 VPs + 2 external advisors) ratifies methodology overlay versions on a monthly, ad-hoc, quarterly, and annual cadence.
§ 02 ── FOUR WAYS METHODOLOGY DILUTES
Four drift vectors at scaling stage.
01
FOUNDER-TO-VP HANDOFF
The handoff is conversational. Neither side gets captured.
Founder methodology was clear in their head. The handoff to VP-level leaders happens conversationally. The VP develops their own methodology. The founder's methodology stays in the founder's head. The VP's methodology stays in the VP's head. Neither is captured. New hires learn the VP's variant, not the founder's.
02
CROSS-FUNCTIONAL COORDINATION DRIFT
Schema divergence surfaces at integration, not at change.
The auth team's API contract changed in a PR two weeks ago. The mobile team's roadmap assumes the old contract. No one filed a coordination ticket. The substrate detects schema divergence within HOURS — via Engineering Steward (Class A · T2 · F6 ceiling) sprint-loop integration plus Strategy Lead (Class A · T2) cross-functional drift detection. Without substrate, the discovery happens at integration start — three weeks later.
03
DECISION PROVENANCE GAPS
Reasoning lives in 1:1s, Slack DMs, dinner conversations.
Strategic decisions get made in 1:1s, Slack DMs, dinner conversations. The reasoning is in someone's memory. Six months later, when the decision is revisited, the reasoning is partial. The substrate captures decisions with full audit-chain context at framing time via Decision Brief Drafter (Class A · T2 · F3 advisory ceiling), audit-emitted by Audit Curator (Class B · T4 · F4 sole-emit · L0 doctrine · NEVER-deletable on critical).
04
METHODOLOGY LEAVES WITH PEOPLE
A VP departs. Three quarters later, the methodology has shifted.
A VP leaves. Their replacement learns “how we do things” through osmosis from the team. Three quarters later, the methodology has shifted. The substrate's L2 overlay preserves methodology in versioned source-of-truth, Council-ratified, customer-owned per MSA § 8.3. Departures don't reset INSTITUTIONAL MEMORY.
FIG.01 — Without capture, methodology fragments at every functional handoff. With the canonical 7-seat Methodology Council + Decision Brief Drafter + Knowledge Curator extended capture, the overlay preserves coherence across handoffs.
FIG.01 — Without capture, methodology fragments at every functional handoff. With the canonical 7-seat Methodology Council + Decision Brief Drafter + Knowledge Curator extended capture, the overlay preserves coherence across handoffs.
§ 03 ── BUILT FOR FUNCTIONAL SATURATION
All seven Class A. All five Class B. Full canonical 12-agent activation.
At Seed-A, the canonical Class A subset activates (typically 5–7 of 7) per the Customization Decision Tree. At Series B/C, all 7 Class A (T2 Drafter) + 5 Class B (T4 Steward) activate — the canonical full operational pattern. Each Class A agent owns its functional area's methodology and operational integration. The Class B architectural agents steward substrate discipline itself: audit chain, L8 floors, κ verification, agent registry, version control.
12 CANONICAL AGENTS · 7 CLASS A + 5 CLASS B · FULL B/C ACTIVATION · TIER CEILING ARCHITECTURALLY ENFORCED
FIG.02 — 12 canonical agents · 7 Class A operational (T2 Drafter ceiling) × 5 Class B architectural (T4 Steward ceiling) · full B/C activation · Year 2 expands to ~25 agents across 6 loops via Phase 3 quarterly accretion.
FIG.02 — 12 canonical agents · 7 Class A operational (T2 Drafter ceiling) × 5 Class B architectural (T4 Steward ceiling) · full B/C activation · Year 2 expands to ~25 agents across 6 loops via Phase 3 quarterly accretion.
CLASS A · 7 CUSTOMER-OPERATIONAL · T2 DRAFTER CEILING
01
Decision Brief Drafter
T2 · F3 advisory ceiling
Cross-functional decision capture · 5-lens scorer (Risk · Customer-Impact · Regulatory · Reversibility · Cost) · diverge-and-reconcile triggers on substantive class · audit-emit via Audit Curator (F4).
02
Risk & Compliance Lead
T2 · F3 advisory ceiling
Per-decision regulatory pathway check · GDPR · SOC 2 (in progress at B/C) · ISO 27001 (scoped) · sector-specific where applicable · employment-law adjacency (F9-adjacent).
03
Customer Operations Lead
T2 · F2 customer-data ceiling
Per-segment customer signal synthesis at scale · sales-call summary · CS health scoring · F2 consent-bounded · drafts customer-facing comms · humans send.
04
Finance Operations Lead
T2 · F1 Money-Movement ceiling
Variance · runway · budget · spend categorization · stage-adapted approval-routing thresholds per the Customization Decision Tree.
05
Knowledge Curator
T2
Decision precedent library · regulatory pathway library · cross-functional knowledge · extended methodology-capture mode at B/C: captures founders + VPs methodology, per-meeting opt-in, consent-bounded.
06
Strategy Lead
T2
Cross-functional strategic synthesis · OKR amendments drafted for Methodology Council ratification · loop-accretion candidate surfacing · roadmap calibration.
07
Engineering Steward
T2 · F6 Code-Shipping ceiling (or F7 if unregulated SaaS)
Sprint-loop integration · PR drafting (no autonomous merge to main) · code-review summary at scale · technical-debt tracking · CI/CD health · operates the “AI software factory” pattern per Diana Hu's YC framing.
CLASS B · 5 ARCHITECTURAL · T4 STEWARD CEILING
08
Audit Curator
T4 · F4 sole-emit · L0 doctrine
Sole-emission per tier · hash-chained · NEVER-deletable on critical · severity-classified.
09
L8-Enforcer
T4 · recursively self-protected
Continuous surveillance of all applicable L8 floors (canonical F1–F6 + optional F7–F9 per pattern) · quarterly self-tests · severity:critical on near-miss · cannot be silenced.
10
Registry Maintainer
T4
Agent registry · capability matrix · integration source health · cross-source schema normalization · weekly sweep.
11
Eval Suite Runner
T4
Canonical eval cases · Cohen's weighted κ ≥ 0.85 verification on scoring + judgment agents · halt authority on κ drift · re-baselined annually by Methodology Council.
12
Version Controller
T4
Methodology overlay version progression (v0.5 → v0.9 → v1.0 → v1.x) · pre-snapshots on every MINOR/MAJOR cycle · replay-test gating on MAJOR migrations · stakeholder concurrence routing.
Cross-functional integration is what scales at B/C. The substrate becomes the institutional integrator the founder used to be — and the routing layer the middle managers used to be. Per Diana Hu's YC framing: AI as the operating system the company runs on, not a tool layered on top.
§ ── METHODOLOGY DECAY · ESTIMATE
What's leaking right now.
Three inputs. A doctrinal estimate of un-captured substantive decisions per quarter. Conservative on purpose.
UN-CAPTURED · ESTIMATE
3,000
SUBSTANTIVE DECISIONS · HORIZON WINDOW
RISK SCORE
60/100
DOCTRINAL · CONSERVATIVE
TIER
MEDIUM
PHASE 0 WORTHWHILE
§ 04 ── THE ENGAGEMENT · STANDARD
Standard timing. Higher operational footprint than Seed-A.
Phase 0 doubles to 2 weeks because the discovery surface includes functional leadership, not just founders. Phase 1 extends to 10 weeks because all 7 Class A activate in sequence + 5 Class B steward throughout. KPI gate at end of Phase 1 — Week 7 canonical per the Series B/C Stage Sub-Framework. Phase 2 carries higher hourly footprint. Phase 3 quarterly residency is commonly engaged.
OPERATIONAL FOOTPRINT · HEAVIER THAN SEED-A · PHASE 0 · 2 WK · FREE → PHASE 1 · 10 WK · BUILD → KPI GATE W7 → PHASE 2 · MONTHLY · 12mo+ → PHASE 3 · QUARTERLY · COMMONLY ENGAGED
FIG.03 — Standard B/C timing. Phase 2 carries the load. Phase 3 quarterly residency commonly engaged. Phase 0 is free.
FIG.03 — Standard B/C timing. Phase 2 carries the load. Phase 3 quarterly residency commonly engaged. Phase 0 is free.
PHASE 0
Discovery interviews with founders + key VPs (Days 1–3). Brand DNA Step 0 research (Days 3–5 · canonical STANDARD depth · 30–40 sources · 8 themes). Methodology Council Workshop (Days 6–8 · 7-seat Council formed · charter ratified · κ baseline lock). Methodology v0.1 capture (15–25 substantive captures · cross-functional scope). Synthetic-data PoC (working prototype · founders + VPs query live). GO/NO-GO at end of Week 2. If NO-GO, customer keeps all artifacts unconditionally per MSA § 8.3.
PHASE 1
All 12 canonical agents activated (7 Class A · T2 + 5 Class B · T4). 6 canonical L8 floors (F1–F6) + 0–3 optional (F7–F9) enforced per the Customization Decision Tree applied to customer pattern (typical B/C: F1–F6 active + F8 if multi-tenant SaaS + F9 if EU/multi-jurisdictional workforce = 7–8 active floors). Methodology capture in extended mode (founders + VPs via Knowledge Curator). Per-functional-leader voice calibration (κ ≥ 0.85 Cohen's weighted · verified by Eval Suite Runner). 8 universal KPI gate at canonical Series B/C Week 7 (+ 0–2 customer-specific K9/K10 if defined at Phase 0). 6-of-8 universal PASS → Phase 2 GO. Internal owner transitioned by W7.
PHASE 2
Higher hourly footprint than Seed-A. 7-seat Methodology Council on monthly + ad-hoc cadence. Quarterly D7 §12 simulated verification (canonical 5 steps · Inspect · Replay · Verify · Cross-check · Escalate). Annual κ re-baselining (Council 6-of-7 quorum). 12-month minimum.
PHASE 3
Evolution work each quarter typical. Loop accretion toward Year-2 ~25 agents across 6 loops via the canonical 6-loop sequence. Pattern migration L2 → L1 candidates surfaced (concurrence-gated per MSA § 8.4). Stage transition to D+ available via the Series B/C → Series D+ Transition Runbook if a funding event triggers.
§ 05 ── ALL 7 CLASS A + 5 CLASS B · CANONICAL FULL B/C OPERATIONAL PATTERN
What “full activation” looks like operationally at B/C scale.
ENGINEERING STEWARD · T2 · F6 (or F7 if unregulated)
Sprint-loop integration · PR drafting (no autonomous merge to main · F6 ceiling, or F7 if customer is unregulated SaaS) · code-review summary at scale · technical-debt tracking · CI/CD health. Operates the “AI software factory” pattern per Diana Hu's TDD-evolution framing: specs and tests drive iteration, humans approve at merge. The 1000x engineer pattern materializes here — not as autonomous coding, but as a substrate that surrounds an engineer with spec-test-iterate capability under F4 audit.
CUSTOMER OPERATIONS LEAD · T2 · F2 ceiling
Multi-segment pipeline · per-segment voice calibration (κ ≥ 0.85 verified by Eval Suite Runner) · sales-call synthesis at scale · campaign attribution · CS health-score tracking · F2 consent-bounded · drafts customer-facing comms · humans send. (GTM + Customer Voice functions consolidated under canonical Customer Operations Lead at v3.1.0.)
STRATEGY LEAD · T2
Cross-functional strategic synthesis · OKR amendments drafted for the 7-seat Methodology Council to ratify · loop-accretion candidate surfacing for the Phase 3 quarterly cadence · roadmap calibration · product methodology capture (canonical: product methodology lives in Strategy Lead + Knowledge Curator, not a standalone “Product Steward”).
FINANCE OPERATIONS LEAD · T2 · F1 ceiling
Variance analysis · runway models · budget tracking · spend categorization. F1 Money-Movement ceiling: stage-adapted approval-routing thresholds per the Customization Decision Tree (specifics discussed at Phase 0). Autonomous money-movement code path does not exist.
RISK & COMPLIANCE LEAD · T2 · F3 advisory ceiling
Per-decision regulatory pathway check. Maps every in-scope decision to applicable frameworks. GDPR (active for EU customer/employee data) · SOC 2 Type II in progress at B/C scale · ISO 27001 (scoped · gap analysis underway) · employment-law adjacency (People functions consolidated under Risk & Compliance Lead + Customer Operations Lead + F9 L8-Enforcer architectural at v3.1.0). F9 (Employee Privacy) typically active at B/C scale due to EU/multi-jurisdictional workforce.
KNOWLEDGE CURATOR · T2
Decision precedent library · regulatory pathway library · cross-functional knowledge. Extended methodology-capture mode at B/C: captures founders + VPs methodology, per-meeting opt-in, consent-bounded. Critical for capturing functional-leadership methodology before dilution. (“Founder Methodology Capture” is a CAPABILITY of Knowledge Curator at v3.1.0, not a standalone agent.)
DECISION BRIEF DRAFTER · T2 · F3 advisory ceiling
Cross-functional decision capture · 5-lens scorer · diverge-and-reconcile triggers on substantive class · audit-emit via Audit Curator (F4 sole-emit). ~47-decision throughput per quarter typical at B/C scale (per anonymized worked-example reference).
CLASS B · 5 ARCHITECTURAL · ALWAYS-ON STEWARDS
Audit Curator (F4 · L0 doctrine) · L8-Enforcer (quarterly self-tests · cannot be silenced) · Registry Maintainer (agent + integration registry · weekly sweep) · Eval Suite Runner (κ ≥ 0.85 halt authority) · Version Controller (overlay v0.5 → v0.9 → v1.0 → v1.x · pre-snapshot enforcement · replay-test MAJOR-cycle gate). All T4 Steward ceiling · recursively self-protected where applicable.
Cross-functional integration is what scales at B/C. The 7 Class A synthesize across functions. The 5 Class B steward substrate discipline itself. The substrate becomes the institutional integrator the founder used to be — and the routing layer the middle managers used to be. Per Jack Dorsey at Block: humans at the edge guiding, agents in the middle routing.
§ 06 ── UNDER THE HOOD · NDA-TIER
B/C-specific depth available pre-engagement.
What's public on this page: 12-agent canonical fabric at B/C full activation · 10-week Phase 1 with W7 KPI gate · 7-seat Methodology Council · 8 universal KPI gate · F1–F6 canonical + optional F7–F9 per pattern · 4-phase engagement. What's available in a 30-min Phase 0 alignment call (informal · pre-NDA):
- 01
Customization Decision Tree applied to YOUR B/C pattern — which subset of F7–F9 optional floors active (F7 if unregulated SaaS · F8 if multi-tenant · F9 if EU employees · combinations per customer).
- 02
Pricing range for Phase 1 fixed-fee + Phase 2 monthly retainer at B/C scale (specifics discussed at Phase 0).
- 03
Brand DNA research-depth recommendation (STANDARD 30–40 sources canonical B/C, or DEEP 40–50+ if multi-jurisdictional regulatory frame).
- 04
First-loop selection — cross-functional decision loop, regulatory operations loop, customer operations loop, etc.
- 05
Methodology Council seat composition — canonical 7 seats at B/C (4–5 internal + 2–3 external/advisor).
- 06
Worked examples at B/C scale (additional anonymized customer studies beyond Ledgerline, available at Phase 0).
- 07
Stage Transition Runbook for B/C → D+ migration when a funding event triggers.
- 08
Pattern migration L2 → L1 mechanics — concurrence-gated · attribution + compensation per MSA § 8.4.
§ ── NOT YOUR STAGE
The substrate is one architecture. The conversation calibrates per stage.
§ END ── BEFORE DILUTION
If you operate a Series B/C company — Phase 0 is two weeks, free. Before functional dilution accelerates. You keep all artifacts either way (per MSA § 8.3).
At a different stage?