1. Essence of a Business Model & Three Outcomes of Billable-Hour Collapse

A business model is how you reliably capture a share of the value you create. How much you capture depends on holding the scarce bottleneck others cannot bypass (Christensen's Law of Conservation of Attractive Profits: when one layer commoditizes, profits shift to an adjacent layer).

First, distinguish "high cost of human time" from "billing by human time". Lawyers bill by the hour, translators by the word, consultants by the person-day, support centers by the seat—all anchoring price on how long a human takes to finish a task. When AI compresses that duration toward zero, the time-billing anchor breaks, leaving three industry outcomes:

1. Incumbents discount and shrink: Hourly rates hold or dip slightly, but total billed hours fall off a cliff.
2. Challengers bundle by outcome (fixed price + liability): Profit shifts from selling time to selling certainty.
3. Value leaks entirely to consumer surplus: Technology creates massive user value, but no supplier holds a defensible bottleneck, collapsing industry revenue.

2. Five Blind Spots Beyond Static Scarcity Maps

1. Task Granularity, Not Industry Labels: Within a single law or consulting firm, routine due diligence collapses into commodity tooling, litigation strategy stays in high-stakes judgment, and partner deal-making remains a trust network.
2. Absolute-Value vs. Relative-Value (Zero-Sum) Markets: Translation and customer support are positive-sum (triggering Jevons paradox: unit price approaches zero, total volume explodes). Trading alpha, ad bidding, and talent wars are zero-sum arms races—when everyone has AI, public-data alpha decays rapidly. Value goes to arms dealers (compute / exclusive ground-truth feeds) or principal capital.
3. The Verification Gap & Lemon Market in "Taste/Judgment": When generating a plausible strategy deck costs near zero, buyers face an ocean of synthetic noise. "Judgment" industries undergo a K-shaped split: mid-tier report factories collapse; only short-cycle measurable outcomes (contingent pricing) or unforgeable public track records and top-tier brand cover survive.
4. Liability Is Actuarial Underwriting + Error Decorrelation: Licensing is regulatory rent with a finite shelf life. Clients pay for liability because the underwriter has capital to pay claims. Single-model bugs create correlated systemic failures; heterogeneous multi-source verification + physical-world grounding reduces error correlation, making risk closer to actuarially insurable (with caps, deductibles, and reinsurance).
5. Living Context = Negative Switching Cost: Static RAG docs are easily migrated. True lock-in lives in accumulated boundary rejections, red-line gates, and private negative-sample evaluation suites.

3. Four Sector Breakdowns: Who Captures What?

4. The 7 Atomic Billing Units & 3-Question Litmus Test

1. Outcome: Contingent fee per resolved ticket, hire, or audited cost saving.
2. Risk Premium: Fixed-price package backed by capped error indemnification and decorrelated verification.
3. Attestation (Verified Receipt): Fee for slashing third-party verification cost via physically grounded and multi-source verified receipts.
4. Workflow Toll: Transaction cut at an unbypassable entry point or closed loop.
5. Exclusive License: Upstream IP rights, proprietary ground-truth feeds, or private red-line benchmarks.
6. Relationship / Retainer: Scarcity of human trust, boardroom political cover, and high-density peer curation.
7. Carry / Equity: Deploying judgment directly into capital rather than selling advice.

The 3-Question Litmus Test to Kill Bad AI Business Models

1. The Bottleneck Question: Once execution is free, who must the customer still pass through to get the final result? Do I hold that node, or will the next foundation model release absorb it natively?
2. The Verification Question: How does the buyer know before paying (or shortly after) that my output isn't synthetic slop? If measurable, am I willing to price on contingency? If unmeasurable, do I have unforgeable reputation cover? (Neither = Lemon Market; kill it.)
3. The Skin-in-the-Game Question: If my signal is truly this accurate, why sell it instead of deploying it myself? Signals that decay with capacity (Alpha) must be traded internally; frameworks and verification substrates that gain network effects with adoption should be opened externally.

5. Falsifiable Public Forecasts (2026–2028 Evaluation Window)

1. Forecast 1 (Legal & Financial Diligence Pricing Shift): By end of 2028, over 50% of routine contract reviews and standard M&A screening engagements will be billed via fixed-fee packages paired with capped error liability/insurance rather than traditional billable hours.
2. Forecast 2 (Mid-Tier Research & Consulting Shakeout): By end of 2028, subscription research boutiques and mid-sized consulting firms lacking either proprietary primary data feeds or contingent/capital-linked pricing will see renewal rates or revenues contract by over 40% relative to 2025.
3. Forecast 3 (Verified Receipts as a Standalone Line Item): By end of 2028, multi-source cross-verified and physically grounded evidence packets (Verified Receipts) will appear as a distinct billable procurement category in enterprise audit, M&A diligence, and high-stakes agent workflows.