Active firm
Each scenario has fully independent state.
Explore mode · Advanced parameters
Override paper-canonical values — NeuroCertify
Each firm in Explore carries its own overrides. Editing here changes the values the framework consumes for NeuroCertify — without touching the other firm. Empty fields stay at paper defaults. The Reset button on the Active-firm bar above clears these too.
CLT employer charges 68-80% (Actual Profit). Imported SaaS adds 25-40% via IRRF + CIDE + PIS/COFINS-imp. Termination ≈ 25-30% of annual base salary.
Converts gross base salary to total employer cost (employer social charges, mandatory benefits, statutory provisions). Drives the absolute size of the labor-substitution premium.
Paper §7.1 + §7.3 — BR 1.7-1.9 (CLT), FR 1.40-1.45 (CDI), US 1.20-1.30 (W-2).
One-time cost at T0 of dismissing the substituted engineers — severance, statutory notice, accrued benefits.
Paper §7.2 — BR 25-30%, FR 30-50%, US 5-15%.
Effective price paid for imported AI services beyond the nominal US$ list price, after import-side withholding, contributions, and VAT/import charges.
Paper §7.4 — BR 1.25-1.40 (IRRF + CIDE + PIS/COFINS-imp); FR 1.20 (TVA-importation 20%, deductible for B2B); US 1.00.
Statutory minimum notice period before dismissal becomes effective. Pushes back the date at which substitution savings begin.
Paper §7.2 — BR 1 month (0.083), FR 2 months (0.167), US 0 (at-will).
Fraction of AI operating expense that can be deducted from the firm's corporate tax base. 1.0 in all three reference regimes (full deductibility).
Paper §7.4 + Appendix D.6 — held at 1.00 across reference regimes; exposed for hypothetical regulatory tightening.
Extra discount-rate premium added to the WACC to price cross-border execution risk (provider concentration, regulatory regime change, data-sovereignty exposure).
Paper §7.3 footnote — BR/FR ≈ 0.5 pp, US ≈ 0.2 pp.
URSSAF total ≈ 40-45% (cadre, 2026). CDI termination 30-50% of annual salary depending on seniority. TVA-importation 20% (deductible B2B).
Converts gross base salary to total employer cost (employer social charges, mandatory benefits, statutory provisions). Drives the absolute size of the labor-substitution premium.
Paper §7.1 + §7.3 — BR 1.7-1.9 (CLT), FR 1.40-1.45 (CDI), US 1.20-1.30 (W-2).
One-time cost at T0 of dismissing the substituted engineers — severance, statutory notice, accrued benefits.
Paper §7.2 — BR 25-30%, FR 30-50%, US 5-15%.
Effective price paid for imported AI services beyond the nominal US$ list price, after import-side withholding, contributions, and VAT/import charges.
Paper §7.4 — BR 1.25-1.40 (IRRF + CIDE + PIS/COFINS-imp); FR 1.20 (TVA-importation 20%, deductible for B2B); US 1.00.
Statutory minimum notice period before dismissal becomes effective. Pushes back the date at which substitution savings begin.
Paper §7.2 — BR 1 month (0.083), FR 2 months (0.167), US 0 (at-will).
Fraction of AI operating expense that can be deducted from the firm's corporate tax base. 1.0 in all three reference regimes (full deductibility).
Paper §7.4 + Appendix D.6 — held at 1.00 across reference regimes; exposed for hypothetical regulatory tightening.
Extra discount-rate premium added to the WACC to price cross-border execution risk (provider concentration, regulatory regime change, data-sovereignty exposure).
Paper §7.3 footnote — BR/FR ≈ 0.5 pp, US ≈ 0.2 pp.
FICA + FUTA + SUTA + workers' comp ≈ 8-15%; voluntary benefits add 15-20%. At-will termination — contractual severance only, ≈ 5-15% of annual salary.
Converts gross base salary to total employer cost (employer social charges, mandatory benefits, statutory provisions). Drives the absolute size of the labor-substitution premium.
Paper §7.1 + §7.3 — BR 1.7-1.9 (CLT), FR 1.40-1.45 (CDI), US 1.20-1.30 (W-2).
One-time cost at T0 of dismissing the substituted engineers — severance, statutory notice, accrued benefits.
Paper §7.2 — BR 25-30%, FR 30-50%, US 5-15%.
Effective price paid for imported AI services beyond the nominal US$ list price, after import-side withholding, contributions, and VAT/import charges.
Paper §7.4 — BR 1.25-1.40 (IRRF + CIDE + PIS/COFINS-imp); FR 1.20 (TVA-importation 20%, deductible for B2B); US 1.00.
Statutory minimum notice period before dismissal becomes effective. Pushes back the date at which substitution savings begin.
Paper §7.2 — BR 1 month (0.083), FR 2 months (0.167), US 0 (at-will).
Fraction of AI operating expense that can be deducted from the firm's corporate tax base. 1.0 in all three reference regimes (full deductibility).
Paper §7.4 + Appendix D.6 — held at 1.00 across reference regimes; exposed for hypothetical regulatory tightening.
Extra discount-rate premium added to the WACC to price cross-border execution risk (provider concentration, regulatory regime change, data-sovereignty exposure).
Paper §7.3 footnote — BR/FR ≈ 0.5 pp, US ≈ 0.2 pp.
Total annual cost (salary + employer charges + benefits) of a senior software engineer. Used for the orchestrator function — the survivor role after the migration.
Paper §7.5 — BR $95k, FR $145k, US $420k (Levels.fyi Q3 2025 + jurisdictional multipliers).
Total annual cost of a mid-tier (substitutable) engineer. Multiplied by the substitution headcount to size the annual gross saving.
Paper §7.5 — BR $60k, FR $95k, US $285k.
Total annual cost (salary + employer charges + benefits) of a senior software engineer. Used for the orchestrator function — the survivor role after the migration.
Paper §7.5 — BR $95k, FR $145k, US $420k (Levels.fyi Q3 2025 + jurisdictional multipliers).
Total annual cost of a mid-tier (substitutable) engineer. Multiplied by the substitution headcount to size the annual gross saving.
Paper §7.5 — BR $60k, FR $95k, US $285k.
Total annual cost (salary + employer charges + benefits) of a senior software engineer. Used for the orchestrator function — the survivor role after the migration.
Paper §7.5 — BR $95k, FR $145k, US $420k (Levels.fyi Q3 2025 + jurisdictional multipliers).
Total annual cost of a mid-tier (substitutable) engineer. Multiplied by the substitution headcount to size the annual gross saving.
Paper §7.5 — BR $60k, FR $95k, US $285k.
Retained engineers per AI orchestrator (the permanent overhead the firm carries after the transition).
Paper §7.5 — default 1:10 (Gartner 2025 + McKinsey 2025 calibration).
AI-skill compensation premium over baseline senior SWE comp. Levels.fyi Q3 2025 calibration.
Paper §7.5 footnote — default 14.2%.
Heavy-use AI tooling cost charged for every retained developer (Cursor / Copilot / inference budget envelope).
Paper §7.5 + Appendix D — default $12 000 (May 2026, heavy-use profile).
Quarters after T0 where the substituted team is partially still on payroll while AI ramps up.
Paper §7.5 — default 3 quarters (T0 → T3).
Quarters the firm pays a retention bonus to remaining senior engineers, to lock in the survivor team during transition.
Paper §7.5 — default 2 quarters (T1 → T2).
Bonus size as a fraction of the departing senior engineer's annual comp, paid per quarter during the retention window.
Paper §7.5 — default 10%.
Pre-T0 calibration period the firm spends scoping AI substitution before any layoffs land. Anchored to Brynjolfsson, Li & Raymond (2025).
Paper §7.5 — default 9 months.
Minimum number of orchestrators per substitution arm — guarantees at least 1 orchestrator on payroll whenever any substitution happens, regardless of headcount math.
Paper §7.5 — default 1 orchestrator/arm.
Each row is one layer's signed risk coefficient α. Positive α contributes to the firm-specific premium (commoditizing); negative α subtracts (protective). L7 is the special case: paper-canonical α₇ = 0, with the K₇ channel adding (1 − K₇) × 0.03 per unit of L7 exposure on top. Setting α₇ to a non-zero value here REPLACES the K₇ channel with the flat value.
Compute / inference / data centre. Slowly commoditizing.
Paper Appendix A.2 — α canonical +0.02.
Frontier model weights + API; commoditizing weakly.
Paper Appendix A.2 — α canonical +0.01.
API + tool integrations. Commoditizing more visibly.
Paper Appendix A.2 — α canonical +0.04.
Retrieval, synthesis, prototyping — the core erosion layer.
Paper Appendix A.2 — α canonical +0.08.
Problem formulation, design choices — partially protective.
Paper Appendix A.2 — α canonical -0.04.
Trust, certification, distribution — most protective.
Paper Appendix A.2 — α canonical -0.06.
Modulator hypothesis (Section 4.1). Paper-canonical α₇ = 0; the K₇ channel substitutes its own (1 − K₇) × 0.03 premium per unit of L7 exposure. Override here REPLACES the K₇ channel with the flat value.
Paper Appendix A.2 — α canonical +0.00.
0 = normal-technology reading (commoditization is business-as-usual); 1 = structural-change reading (post-AI window is a regime shift). The dial modulates the framing language in the multi-audience reports.
Paper §B.5 (Macro Integration Proposal) — default 0.50.
Paper §B.5 + Appendix E (Carta Q3 2025) — affects the seed reference line on funding-stage figures and the report framing only.
Each layer has two paper-canonical constants — its annualised commoditization velocity (Δ substitutability per year) and its 2026 substitutability level. Editing either reshapes the seven-layer trajectory chart on /demo/layers and /value/deep-dive/layers directly. L7 (cross-border knowledge regime) is the §4.1 modulator, not a Figure-1 layer, so it is not editable here.
Paper Figure 1 — velocity sign drives the chip colour (green commoditizing, orange weakly, red anti-commoditizing).
Three reference regimes anchor the K₇ axis on every cross-border figure. Editing a regime's K_coefficient changes the slider reference labels and the resolved values returned by /layers/knowledge-regimes. cross_border_friction is the global multiplier the framework applies to substitution potential when the target and acquirer are in different blocs.
Global multiplier (0..1). Applied as 1 − cbf to substitution potential when target/acquirer blocs differ — drives the cross-border friction term in crossborder_acquisition_friction.
Paper §4.1 — default 0.30.
Pre-decoupling baseline. Approximate state circa 2018-2020.
The regime's anchor value for K₇ (cross-border knowledge integration). Reference points for the K₇ slider on /demo/layers and /value/deep-dive/layers.
Paper §4.1 — globalized 1.00 / current 0.70 / fragmented 0.40.
Multiplier applied to Layer-4 substitutability when the regime is in effect — lower in fragmented regimes (smaller corpora, less knowledge sharing).
Paper §4.1 / `apply_knowledge_regime_modulation` — defaults 1.00 / 0.70 / 0.40.
Divisor on Layer-5 judgment value. Lower factor ⇒ higher judgment-value premium (scarcer human judgment in fragmented regimes).
Paper §4.1 — defaults 1.00 / 0.85 / 0.55. Floor 0.10 in the framework.
Current estimated regime — partial fragmentation. Illustrative, not estimated.
The regime's anchor value for K₇ (cross-border knowledge integration). Reference points for the K₇ slider on /demo/layers and /value/deep-dive/layers.
Paper §4.1 — globalized 1.00 / current 0.70 / fragmented 0.40.
Multiplier applied to Layer-4 substitutability when the regime is in effect — lower in fragmented regimes (smaller corpora, less knowledge sharing).
Paper §4.1 / `apply_knowledge_regime_modulation` — defaults 1.00 / 0.70 / 0.40.
Divisor on Layer-5 judgment value. Lower factor ⇒ higher judgment-value premium (scarcer human judgment in fragmented regimes).
Paper §4.1 — defaults 1.00 / 0.85 / 0.55. Floor 0.10 in the framework.
Hypothetical 2030 severe-fragmentation counterfactual.
The regime's anchor value for K₇ (cross-border knowledge integration). Reference points for the K₇ slider on /demo/layers and /value/deep-dive/layers.
Paper §4.1 — globalized 1.00 / current 0.70 / fragmented 0.40.
Multiplier applied to Layer-4 substitutability when the regime is in effect — lower in fragmented regimes (smaller corpora, less knowledge sharing).
Paper §4.1 / `apply_knowledge_regime_modulation` — defaults 1.00 / 0.70 / 0.40.
Divisor on Layer-5 judgment value. Lower factor ⇒ higher judgment-value premium (scarcer human judgment in fragmented regimes).
Paper §4.1 — defaults 1.00 / 0.85 / 0.55. Floor 0.10 in the framework.
Five-year fiscal projection across the three reference jurisdictions (Figure D.8). The headline number per bloc is lost_social_charges + ai_token_export − compensating_tax_gain — positive means the state loses revenue. Edits propagate live to the Appendix-D panel's fiscal-blocs chart via /appendix-d/fiscal/projections/resolve.
Length of the cumulative-impact projection.
Scaling factor from the single-firm reference to the sector aggregate.
Share of margin gain the HQ captures via transfer pricing.
Share of margin gain the local subsidiary retains via transfer pricing.
Paper Appendix D.6 — defaults: horizon 5y, multiplier ×6, TP parent 70% / subsidiary 15%.
Combined statutory corporate tax — applied to the higher operating margin enabled by AI substitution.
Employer-side social charges as a fraction of gross salary. Eliminated when the role is substituted.
Cumulative 5-year revenue loss from substituted employer charges.
Tax base migrated to a foreign AI provider — positive = exported (loss), negative = captured domestically (gain).
Cumulative 5-year gain from corporate tax on the higher operating margin.
Combined statutory corporate tax — applied to the higher operating margin enabled by AI substitution.
Employer-side social charges as a fraction of gross salary. Eliminated when the role is substituted.
Cumulative 5-year revenue loss from substituted employer charges.
Tax base migrated to a foreign AI provider — positive = exported (loss), negative = captured domestically (gain).
Cumulative 5-year gain from corporate tax on the higher operating margin.
Combined statutory corporate tax — applied to the higher operating margin enabled by AI substitution.
Employer-side social charges as a fraction of gross salary. Eliminated when the role is substituted.
Cumulative 5-year revenue loss from substituted employer charges.
Tax base migrated to a foreign AI provider — positive = exported (loss), negative = captured domestically (gain).
Cumulative 5-year gain from corporate tax on the higher operating margin.
Multiplier on the firm's Layer-6 share inside the fragility index. Larger values give institutional embedding more weight, pulling more firms into the resilient zone.
Paper §E.5 — default 1.5 (illustrative).
Upper bound of the resilient zone. Firms with fragility index below this value plot green on the map.
Paper §E.5 — default −0.10.
Lower bound of the fragile zone. Firms with fragility index above this value plot red on the map. Must be strictly greater than the resilient threshold to leave a borderline band.
Paper §E.5 — default +0.10.
Carta State of Private Markets Q3 2025 medians for the five reference stages. Edits propagate live to the Reference funding stages card on /demo/appendix-e-dynamic via /appendix-e/funding-stages/resolve.
AI-native firms raise rounds this fraction smaller than legacy peers.
Median dilution per round for legacy (non-AI-native) firms.
Median dilution per round for AI-native firms — smaller rounds drive smaller dilution.
Paper Appendix E — Carta Q3 2025 defaults: 35% reduction, 22% legacy dilution, 17.5% AI-native dilution.
Median funding round size at this stage (Carta Q3 2025).
Median pre-money valuation at this stage (Carta Q3 2025).
Typical equity dilution paid by founders/existing holders at this stage.
Median funding round size at this stage (Carta Q3 2025).
Median pre-money valuation at this stage (Carta Q3 2025).
Typical equity dilution paid by founders/existing holders at this stage.
Median funding round size at this stage (Carta Q3 2025).
Median pre-money valuation at this stage (Carta Q3 2025).
Typical equity dilution paid by founders/existing holders at this stage.
Median funding round size at this stage (Carta Q3 2025).
Median pre-money valuation at this stage (Carta Q3 2025).
Typical equity dilution paid by founders/existing holders at this stage.
Median funding round size at this stage (Carta Q3 2025).
Median pre-money valuation at this stage (Carta Q3 2025).
Typical equity dilution paid by founders/existing holders at this stage.
Pre-valley revenue-retreat factor. Default 1.00 = no retreat outside the valley.
Paper §B.2.6 Eq B.14 — default 1.00.
Post-valley revenue-retreat factor in the unified-lambda variant. < 1.0 = permanent margin compression.
Paper §B.2.6 (unified) — default 1.00 in literal Eq B.15.
Systematic share of Layer-4 risk already carried by the Phase-2 beta jump. α_4_adj = α_4 − α_4_sys in the dual-channel path only.
Paper §B.2.6 Eq B.12 — default 0.03.
λ_2V_phase2 = clamp(1 − k_L4 · L4_share + k_L6 · L6_share, lower, upper).
Layer-4 weight in the Phase-2 retreat. Higher = more substitution risk.
Paper §B.2.6 — default 0.55.
Layer-6 protection weight. Higher = institutional embedding offsets more of the retreat.
Paper §B.2.6 — default 0.40.
Floor of the Phase-2 clamp.
Paper §B.2.6 — default 0.50.
Ceiling of the Phase-2 clamp.
Paper §B.2.6 — default 1.00.
λ_2V_phase3 = clamp(1 − k_L4_p3 · L4_share + k_L6_p3 · L6_share, lower, upper). Phase-3 coefficients are calibrated separately — k_L4_p3 > k_L4 captures that permanent damage exceeds the transient dip.
Layer-4 weight in the Phase-3 (permanent) retreat. Default higher than k_L4.
Paper §B.2.6 unified — default 0.85.
Layer-6 protection in the Phase-3 retreat — Layer-6 advantage persists.
Paper §B.2.6 unified — default 0.40 (same as k_L6).
Floor of the Phase-3 clamp.
Paper §B.2.6 unified — default 0.50.
Ceiling of the Phase-3 clamp — even Layer-6 dominant firms suffer some residual.
Paper §B.2.6 unified — default 0.95.
Phase-2 retreat for the Layer-6-rich firm — mild.
Paper §B.2.6 — default 0.95.
Phase-2 retreat for the Layer-4-heavy firm — severe.
Paper §B.2.6 — default 0.70.
Permanent post-valley compression for the Layer-6-rich firm.
Paper §B.2.6 unified — default 0.95.
Permanent post-valley compression for the Layer-4-heavy firm.
Paper §B.2.6 unified — default 0.57.
Distribution + half-width + clamp bounds for the λ_2V_phase2 and α₄_sys draws (consumed by Sprint 3 MC).
± around per-firm calibration.
Paper §B.2.6 MC — default 0.10.
Draw floor.
Paper §B.2.6 MC — default 0.50.
Draw ceiling.
Paper §B.2.6 MC — default 1.00.
± around α₄_sys.
Paper §B.2.6 MC — default 0.015.
α₄_sys cannot go below 0.
Paper §B.2.6 MC — default 0.00.
α₄_sys cannot exceed α₄ itself.
Paper §B.2.6 MC — default 0.08.
Each TRL maps to a percentage-point premium that the Layered (A) DCF adds on top of base CAPM. The schedule decays from +16 pp at TRL 1 (basic principles) down to 0 pp at TRL 9 (proven in operations) — calibrated to Equidam (2025) and Hectelion (2025). Editing here re-anchors every per-firm discount-rate calculation.
Paper Appendix A — TRL discount premium schedule (Equidam 2025, Hectelion 2025).
Five scalars govern when the classical key-person discount flips into the inverted-discount premium. They drive the Damodaran baseline used by the four-path comparison and the inverted- discount heatmap on /demo/inverted-discount.
Legacy Damodaran penalty for firms whose value depends on a small, indispensable team. Applied when the inverted regime does NOT kick in.
Paper §7 — default 17.5%.
Once a firm's Layer-4 (codified work) share crosses this fraction, the key-person becomes an asset (orchestrator) and the discount inverts into a premium.
Paper §7 + Appendix A — default 0.55.
Upper bound on the upside the inverted-discount regime can produce. Caps how aggressively the model rewards Layer-4-heavy firms.
Paper §7 — default 15%.
Below this AI substitution potential, the classical discount still wins regardless of Layer-4 share — there's no orchestrator to extract value from.
Paper §7 — default 30%.
Gordon-model terminal growth used by the full Damodaran DCF anchor. The four-path inversion uses it as the long-run reference.
Paper §7 + Appendix A — default 3%.
Global fallback defaults for the two-phase B reformulation. The per-firm trajectories in Configuration override these when set; editing here re-anchors any firm that doesn't carry its own phase trajectory. β jumps in Phase 2 capture the second-valley risk spike (post-AI); D/E and kd spreads track the financing profile across phases.
Paper Appendix B — defaults: phase boundaries Y2 / Y4, tax 25%.
Paper Appendix B (Eqs B.3-B.11) — Phase 2 β jump captures the post-AI second-valley systematic risk; Phase 3 settles at a new normal.
The hype-cycle curve has two shapes — Classical (single peak → trough → plateau) and Post-genAI (two peaks separated by a commoditization valley). The death-valley templates govern cash trajectories: how runway shrinks, how revenue ramps, when refinancing arrives. Edit any of these and the curves on /demo/hype-cycle reshape directly.
Paper §6.5 — Figure 6.5 (Gartner-style hype curve, single vs double valley) + death-valley cash trajectory templates.
Two sub-blocks: the double-threshold (Fig G.1) for AI substitution in regulated small firms — every decision must cross both an economic break-even AND a regulatory (XAI compliance) floor; and the XAI capacity gap (Fig G.2) across blocs A and B under three K₇ regimes. Edits propagate to /demo/appendix-g.
L4 share × AI sub potential × loaded SWE cost differential.
Economic break-even — minimum savings to cover the orchestrator.
Regulatory break-even — minimum to meet explainable-AI infrastructure requirements.
Annual growth factors per K₇ regime — bloc A is the leading bloc, bloc B the lagging. Lower K₇ ⇒ wider gap.
Paper Appendix G — Figures G.1 (double threshold) and G.2 (XAI capacity gap across blocs).
Capex-sensitivity decay exponents reshape the upstream chart on /demo/appendix-f: training-capex (cumulative, anti-commoditizing) decays faster than inference- capex (marginal) as financing tightens. The exposure matrix is the source for the upstream-chain heat-map — each cell is an intensity score 0–3 (3 = predominant, 0 = none).
Higher = steeper decay of training-capex as credit tightens.
Lower = gentler decay; inference-capex is more marginal.
Tightness fraction below which credit is considered loose.
Tightness fraction above which credit is considered tight.
Intensity per cell: 0 = none, 1 = marginal, 2 = secondary, 3 = predominant. Edits propagate to the upstream heat-map.
| Category | L1 train | L1 infer | L2 | L3 | L4 | L5 | L6 |
|---|---|---|---|---|---|---|---|
| Foundry pure-plays (TSMC, GF) | |||||||
| Training silicon (NVIDIA H/B, AMD MI) | |||||||
| Inference & edge silicon (ASICs, NPUs) | |||||||
| Memory & HBM (Micron, SK Hynix, Samsung) | |||||||
| Hyperscalers (AWS, Azure, GCP) | |||||||
| Frontier labs (Anthropic, OpenAI, DeepMind) | |||||||
| AI-tooling platforms (Cursor, Lovable, Decagon) |
Double-click any edited cell to revert it to paper.
Paper Appendix F — upstream-chain table + Figure F.3.
Two thesis profiles share most fields. The AI-aware variant (§6.1) pivots weight from team_quality toward hypothesis_quality (L5) and institutional_embedding (L6) — the two layers the paper identifies as anti-commoditizing under AI substitution.
Investor's required IRR — drives the VC-method valuation.
Years from investment to exit.
Normalised score above which the thesis recommends 'fund'.
Median dilution per round in the absence of stage-specific data.
Multiple of fair price above which investor walks away.
Paper §3 (Investor scoring) + §6.1 (AI-aware thesis).
Each CV controls how wide the log-normal perturbation is for one input dimension. Tighter CVs narrow the P10-P90 envelope; broader CVs widen it. Used by every Monte Carlo run across the site (Configuration, Reports figures section, Sensitivity step).
Log-normal coefficient of variation applied to team-size draws.
Log-normal CV on monthly burn draws.
Log-normal CV on AI substitution potential draws.
Log-normal CV on revenue-multiple draws.
Log-normal CV on per-layer velocity draws.
Additive noise on substitutability_2026 per layer.
Paper Appendix B + §B.5 — Monte Carlo perturbation spec.
Each non-DCF method (Berkus, VC, Damodaran classical) carries a low/high band factor that brackets the point estimate. The comparable-multiples method uses a baseline revenue multiple and a post-genAI volatility scalar to construct its spread.
Lower bracket for the Berkus pre-revenue method.
Upper bracket for the Berkus pre-revenue method.
Lower bracket for the VC method (target IRR backwards).
Upper bracket for the VC method.
Lower bracket on the Damodaran classical point estimate.
Upper bracket on the Damodaran classical point estimate.
Default revenue multiple used by the comparable-multiples method.
Spread scalar applied to the baseline multiple to construct the band.
Weight applied to the Berkus 'Prototype' dimension since genAI made prototypes cheap-to-produce. 1.0 = classical Berkus; 0.55 = paper §6.1 default (45% decay).
Paper §7 — multi-method reconciliation (Berkus, VC, Damodaran, comparable multiples). Berkus 2026 decay anchored at paper §6.1.
Two scalars in valuation_layered that modulate how the per-layer α coefficients scale with substitution potential (L4) and cross-border friction (L7). They are NOT per-layer values — they sit alongside the seven α coefficients and amplify two of them.
Additive base in the L4 amplifier: effective L4 premium = α₄ × (base + s), where s is AI substitution potential. Base 0.5 floors the amplifier at 0.5 (s=0) and grows linearly to 1.5 (s=1).
Paper §6.4 + Eq. B.13 — default 0.50.
Risk-premium adder per unit of (1 − K₇) in the L7 cross-border modulator. Effective L7 premium = (1 − K₇) × value. A fragmented world (K₇=0) adds the full premium; a globalized one (K₇=1) adds nothing.
Paper §4.1 — default 0.03 (3 pp).
Per-quarter fraction of structural saving captured. The default curve is slow until Q5 (assessment + dual-operation overhead absorb the gain) then sigmoidal lift to steady state at Q10. Each anchor is independent — edits do NOT preserve monotonicity.
Paper §7.5 + Brynjolfsson, Li & Raymond (2025) — sigmoidal learning-curve anchors. YAML: migration_dynamics.learning_curve.
Compounded year-over-year multiplier applied to the per-jurisdiction ai_service_overhead to produce an effective AI-cost trajectory. Year 0 = 1.0; Year y = multipliery. The paper's actual simulation holds prices stable (multiplier = 1.0); the paper-cited reference is ~0.10/year (Alexandre 2026).
Multiplier applied each year. 1.0 = stable; 0.75 = 25%/yr decline; 0.10 = order-of-magnitude/yr decline (paper-cited reference).
Paper §10 Limitation 4 — default 1.0 (snapshot at 2026 prices).
Length of the sweep window. The trajectory is multiplier compounded over Year 0 through Year N.
Same picture, finer brush. Set what you actually pay per million tokens and roughly how many tokens land on a developer's account each month — the simulator does the multiplication and plugs the result back into migration, payback and break-even exactly as it would with a flat dollar figure.
What your provider charges for a million input tokens. In mid-2026 Anthropic Sonnet 4.6 lists at $3, Opus 4.7 around $15, Haiku 4.5 under $1; a self-hosted open-weight stack lands closer to twenty cents.
Same shape for what the model writes back. Output usually runs four to five times input across the major catalogues — Sonnet around $15, Opus $75, Haiku $4, self-hosted under $1.
Roughly how much context a developer pushes through the model in a month. Heavy daily use of an agentic assistant adds up faster than people expect; the cheaper tiers often invite even larger context windows.
Code generation, planning notes and refactor diffs put together rarely top a fifth of what the model had to read — pricier tiers can shift that ratio when the workload leans on long deliberation.
A free-form note so a future you — or anyone looking at the exported scenario — knows which contract these numbers came from. Doesn't change any calculation.
Three substitution scenarios bracket the streaming-case sensitivity. The central case (60%) is what the Figure D.7 baseline uses. Cross-bloc friction (30%) reduces effective substitution when orchestrator and target are in different blocs.
Bottom of the plausible substitution range — low takeup, partial substitution.
Paper §D.1.5 — default 40%.
Central case used in the streaming-case Figure D.7 baseline.
Paper §D.1.5 — default 60%.
Top of the plausible substitution range — full Layer-4 codified takeup.
Paper §D.1.5 — default 78%.
Reduction applied to effective substitution when the AI orchestrator's bloc differs from the target market's. Knowledge regime friction (§4.1) translated into a streaming-case scalar.
Paper §D.1.6 — default 30%.
Paper-cited reference anchors for the 50-engineer / 60%-substitution firm used in Figures 9 and 11. Editing these does NOT change the migration cashflow — they're surfaced for comparison against the resolved per-jurisdiction migration result.
Headcount of the reference firm before substitution.
Fraction of the team that's Layer-4 codified and replaceable.
| Jurisdiction | Break-even (months) | Cum. 5-y gain (USD M) |
|---|---|---|
| Brazil (CLT) | paper 21 | paper 3.7 |
| France (CDI) | paper 21 | paper 7.8 |
| United States (W-2) | paper 16 | paper 18.4 |
Paper §7.5 + Figures 9 & 11. YAML: migration_dynamics.reference_firm.
Each value is a fraction of the standard plan price ($15.49). L4 lines (engineering, support) are the ones AI substitution operates on. The sum should be 1.00 — if it isn't, the streaming-case scenarios will drift off the baseline.
Σ = 1.000Content rights + originals — Layer-6 institutional, not AI-substitutable.
L6 — untouchable
Platform engineering — Layer-4 codified work, the core AI-substitutable line.
L4 — substitutable
Support agents — Layer-4, AI-substitutable via tier-1 chat/voice agents.
L4 — substitutable
Inference/serving compute — Layer-1, partially compressible by efficiency gains.
L1 — partially compressible
Acquisition + brand — partially codifiable (ad ops) but not core to thesis.
ambiguous
Back-office, legal, finance — modest substitution potential.
ambiguous
The firm's take. Mathematical residual — increases as L4 lines compress.
residual
Per firm × entry stage, the multiplier the investor expects at the 10-year horizon. The legacy → AI-native delta per stage is the Carta AI premium — shown live in the bottom row.
| Firm profile | Pre-seed | Seed | Series A |
|---|---|---|---|
| NeuroCertify — legacy | paper 10.0× | paper 6.5× | paper 4.0× |
| NeuroCertify — AI-native | paper 14.0× | paper 9.0× | paper 5.5× |
| DataFlow — legacy | paper 7.5× | paper 4.5× | n/a |
| DataFlow — AI-native | paper 10.0× | paper 6.0× | n/a |
| Carta AI premium (NC) | +4.0× | +2.5× | +1.5× |
| Carta AI premium (DF) | +2.5× | +1.5× | — |
Paper §E.5 + Appendix A.4. YAML: funding_stages_carta.expected_multiple.
Anchor constants for the inverted-discount heatmap on /demo/inverted-discount. They control the dollar-axis scaling and the color-scale bounds — they don't change the underlying percentage-point premium math.
Reference enterprise value the inverted-discount heatmap scales against. Larger EV = larger dollar deltas for the same percentage-point premium.
Paper sweeps — default $100M.
Lower bound of the color scale on the inverted heatmap (negative = classical penalty side).
Paper sweeps — default −15 pp.
Upper bound of the color scale on the inverted heatmap (positive = inverted premium side).
Paper sweeps — default +20 pp.