cavendish
SignalContestedgate · EconomicsNext · 3–7 years×2 sightings

Post Economy

When cognition is priced per token rather than per hour, the consultancy business model inverts: the scarce input becomes judgement about what to ask and what to trust, and billing by effort becomes indefensible within the decade.

Clustered only. No lab work behind it. Cannot be cited.

Join with…

Confidence

35%human-committed

Expiry

63duntil review · 5 Nov 2026

Lead time

not yet mainstream · opened 12 Jan 2026

Ownership

RMRohan Mehtasponsor · prospective vertical

Where it is

This field was placed on the whiteboard twice, at Next and at Distant, and the horizon itself is contested — that is why it is in the contested state without an owner. The conviction behind it is the lab director's: the price of a unit of cognition has fallen roughly 10× a year for three years and shows no floor, and no service business built on hourly labour survives that curve unchanged. The evidence is real on price and thin on consequence. Token prices are measurable and falling; the substitution of tokens for billable hours is visible in our own delivery telemetry; what a post-labour-pricing consultancy looks like is not visible anywhere yet. The Next placement says the business-model shift starts inside five years; the Distant placement says the macro 'post economy' — abundant cognition changing what work is — is a different question and much further out. Both are recorded.

Why a Quantium decision hinges on it

Quantium sells analysis by the engagement, priced on people-time. If a client can buy the same analysis for tokens, the firm either moves up to judgement and accountability or competes on price with a model garden. That is the exec question behind this field and the reason it was placed by conviction rather than waiting for a signal that would arrive too late to act on.

Field attributes

StateContested
GateEconomics · possible, not yet affordable
OriginConviction
Measurablepartial
Audience · TLPexec
Horizonnext
Opened12 Jan 2026
Mainstreamnot yet
Last validated5 Aug 2026
Sightings2

Position

What is demonstrated, what is hype, what would have to be true.

The shape every position request answers. Signal-tier fields carry a draft; assessed and tested fields carry a validated one.

What is demonstrated
  • 01Frontier-class $/Mtok has fallen roughly 10× a year since 2023 at constant capability; the curve has not flattened (sa-cost-per-mtok tracks the current number).
  • 02In the lab's own delivery telemetry, tokens now do work that was 20–30% of junior analyst hours on standard engagements.
  • 03At least two consultancies have publicly moved parts of their analytics work to outcome or subscription pricing, citing model costs.
What is hype
  • 01'The end of work' essays that treat a price curve as a social outcome.
  • 02Headcount projections extrapolated from token prices without accounting for the judgement and accountability roles that do not get cheaper.
  • 03Consultancy 'AI-native' rebrands that keep hourly billing underneath.
What would have to be true
  • 01Token cost per completed analytical task falling below 1% of the labour cost it replaces — the point at which price, not capability, is the client's argument.
  • 02A measured substitution rather than augmentation in a Nightingale run: fewer hours billed for the same output, not the same hours producing more.
  • 03A client willing to buy judgement and accountability separately from analysis; today they are bundled in the hourly rate.
What we would do
  • 01Keep the two-horizon placement explicit until a Nightingale measurement settles which one this is; do not elect an owner until it does.
  • 02Instrument the lab's own delivery for tokens-per-task against hours-per-task, as the cheapest available measurement of the substitution rate.
  • 03Brief the exec sponsor quarterly on the price curve and the substitution rate; this is a board question, not a lab one.

Signals · 9 in this cluster

What the cluster is made of.

Every item carries its source, tier and sightings. Detector-found signal sits beside human drops; downstream they are indistinguishable except by provenance.

band 1 · bleeding edgeband 2 · early adoptionband 3 · demand
10.4×
annual decline
Benchmark·band 1Assessed

Constant-capability inference price index: 10.4× annual decline, three-year trailing

Argus tracks $/Mtok for the cheapest model that clears a fixed capability bar on the lab's evals. The annual decline has been between 8× and 12× each year since 2023. No floor visible. The field's load-bearing number.

extracted claimInference price at constant capability falls roughly 10× a year with no observed floor.
Lab · Argus price tracker1 Aug 2026
detector · bleeding edge
Finding·band 1Tried

Delivery telemetry: tokens substitute for 20–30% of junior analyst hours on standard engagements

Tokens-per-task and hours-per-task on six standard engagement patterns, two quarters. The substituted share is junior hours on data preparation and first-draft analysis. Senior hours unchanged. Tried tier: telemetry, not a preregistered run.

extracted claimTokens substitute for 20–30% of junior analyst hours; senior hours are unchanged.
20–30%junior hours substituted
Lab · Nightingale phase 1 · Jun Park24 Jul 2026
JPdropped
Announcement·band 1Signal

Two consultancies move analytics work to outcome and subscription pricing, citing model costs

A global consultancy and a specialist analytics firm both announced outcome pricing for standardised analytics. Both kept day rates for advisory. Argus positioning delta: the shift is on the substituted share, exactly as claim 4 says.

Competitor announcements9 Jul 2026
AWdropped 2
Post·band 2Signal

'Cognition is free now. What are you for?'

The canonical 'post economy' essay: argues abundant cognition ends effort-based pricing everywhere within five years. Widely shared by the sponsor. No data; kept as the thesis statement at its most extreme.

Substack · A well-followed essayist on AI and work29 Jun 2026
RM?dropped 4
Client question·band 3Signal

'Why am I paying day rates for something your model does in an hour?'

Asked by a procurement lead during a renewal. The first time the substitution argument arrived from the client side. Logged as the field's only client-side signal; Engel cannot measure demand for the firm's own repricing.

Engel · banking engagement14 May 2026
CDdropped
Paper·band 2Signal

Jevons in the Knowledge Economy: Demand Response to Falling Analytical Costs

Cross-industry study: every prior fall in the cost of analysis increased demand for analysts. Argues the same will hold for tokens until the output, not the input, is commoditised. The red team's paper.

NBER working paper · Labour economists6 May 2026
CDdropped 2
Analyst·band 3Signal

Analyst: consulting revenue exposed to token substitution 'between 15% and 40%' depending on definition

A range wide enough to be useless, which is itself the finding: nobody has a settled definition of the substitutable share. Our telemetry number sits inside it.

15–40%exposed revenue
Industry analyst21 Apr 2026
detector · demand
Paper·band 1Signal

The Price of a Thought: Inference Cost Curves 2022–2026

Decomposes the price decline into hardware, algorithmic and competition components. Finds algorithmic gains dominate and show no sign of exhaustion. The independent corroboration of the tracker.

extracted claimAlgorithmic efficiency, not hardware, drives most of the inference price decline and is not exhausted.
arxiv.org · Economists and an infra researcher18 Mar 2026
PRAWdropped 3
Talk·band 3Signal

'You cannot bill for a token. You can bill for being wrong.'

A former Big Four partner argues the residual in consulting is accountability — the willingness to be named when the answer is wrong — and that this does not get cheaper. The clearest statement of what the firm would price instead of hours.

Consulting industry conference5 Mar 2026
RMdropped 2
Seen something that belongs here?Under fifteen seconds, or it will not be used.

Claims · 4 supporting, 1 refuting

The atoms.

A document cannot go stale; an assertion can. Claims are immutable and stamped with the extractor that produced them, so staleness, diffs and the graveyard operate at claim level.

Frontier-class inference price at constant capability has fallen roughly 10× a year for three years with no observed floor.

Assessedc-post-economy-1dalton-0.45 Aug 2026Lab · Argus price tracker, arxiv.org
86%

Consultancies that move to outcome pricing do so on the parts of their work where tokens already substitute for hours, not across the firm.

Assessedc-post-economy-4dalton-0.45 Aug 2026Competitor announcements, Industry analyst
64%

In the lab's own delivery, tokens now substitute for 20–30% of junior analyst hours on standard engagements.

Triedc-post-economy-2dalton-0.428 Jul 2026Lab · Nightingale phase 1
60%

Judgement and accountability do not fall on the token price curve; they are the residual the firm prices when analysis does not.

Assessedc-post-economy-5dalton-0.319 May 2026Engel · banking engagement, Consulting industry conference, NBER working paper
55%

Abundant cognition reduces aggregate demand for analytical labour within the Next horizon.

Signalc-post-economy-3dalton-0.45 Aug 2026Substack, NBER working paper
30%

Position history · the diff is the product

2 validation runs against a fixed brief. Confidence 30% → 35%.

runs compare claim sets, never prose
What we said · run 2

Own-delivery substitution measured at 20–30% of junior hours. Outcome pricing appearing at competitors on the substituted share only. Jevons counter-case unanswered; horizon still contested.

35%
Changed since run 1
  • Frontier-class inference price at constant capability has fallen roughly 10× a year for three years with no observed floor.
  • In the lab's own delivery, tokens now substitute for 20–30% of junior analyst hours on standard engagements.
  • Consultancies that move to outcome pricing do so on the parts of their work where tokens already substitute for hours, not across the firm.
  • c-post-economy-3 ↓ 0.38 → 0.3
Positions are superseded, never edited. The prediction record is worthless if it can be quietly revised.Crystal ball

Scoring · ordinal bands

Agents propose. A named human commits.

Uncommitted scores are visibly marked and never leave the building. Bands, not point estimates — false precision is the tell that a number was generated rather than derived.

Impact

committed · AW
high

It is the firm's business model. Conviction placement; scored as such.

Timeline

committed · AW
18mo–4yr

For the business-model shift. The Distant placement (macro post-economy) is a separate 4yr+ band held by the sponsor and not committed here.

TAM

agent-estimated
>$10B

Agent-estimated on global consulting revenue exposed to token substitution. Uncommitted; the number is not the point.

Cost

committed · JP
low

Instrumenting our own delivery is a telemetry change, not an experiment.

Cost of being wrong

committed · RM
high

Wrong late means competing on price with a model garden. Wrong early means repricing a business that did not need it yet.

Demand

agent-estimated
not-measurable-here

Clients do not ask about the consultancy's business model; the demand is the firm's own and Engel does not measure it. Agent-drafted.

Workforce readiness

agent-estimated
low

No delivery team prices on outcome; no partner has sold judgement unbundled from analysis. Agent-estimated.

Relevance · per vertical

Why it matters here, or explicitly does not.

Ranking is per vertical, not global. Sector owners commit notes against agent drafts.

Cross-sector
relevant

The field is about the firm, not a client vertical. Every engagement is exposed to the same curve.

Mechanism · Tokens-per-task versus hours-per-task in delivery telemetry; outcome pricing on the substituted share.

RM committed by Rohan Mehtacommitted
Banking
watch

Banks are the clients most likely to buy analysis from a model garden directly; also the most likely to keep paying for accountability.

Mechanism · Watch for the first bank RFP that specifies token budget rather than day rate.

CD committed by Claire Duboiscommitted
Government
not-relevant

Public-sector procurement prices on day rates by panel rule; the business-model shift is blocked by procurement regardless of the curve on this horizon.

Mechanism · None until panel arrangements change.

Agent draft · awaiting a sector owneragent-estimated

Red team · the strongest case against

The strongest case against: Jevons. Cheaper cognition has, so far, increased demand for analytical work rather than reduced it, and the firm's own telemetry shows more analysis per engagement, not fewer hours. A price curve on an input says nothing about the price of the output when demand is elastic. The conviction may be confusing the cost of tokens with the value of answers, and the two-horizon placement is a way of never being wrong.

  • Every prior automation of analytical work (spreadsheets, BI, cloud) increased analyst headcount. The base rate is against the thesis.
  • The lab's substitution figure is on junior hours, which were never the margin; the pricing conversation is about senior judgement, which the curve does not touch.
  • A field placed at two horizons with no owner cannot be falsified; it should either be elected at one horizon or dissolved.
Stored permanently alongside the thesis. Sources are correlated; without an adversary, synthesis converges on consensus and calls it insight.thesis in doubt

Source diversity

  • Economics research30%
  • Consulting industry25%
  • Commentary15%
  • Internal / Engel30%

A field supported by one epistemic community is a flag, not a finding.

Cross-pollination · typed joins

Connected, not merely similar.

Enabling, compounding, substituting, blocking. A satisfied dependency trigger is a far stronger signal than semantic proximity.

Trigger · Frontier-class $/Mtok falls to the point where token cost per completed analytical task is under 1% of the labour cost it replaces on a standard engagement; Argus computes this quarterly from sa-cost-per-mtok and the lab's tokens-per-task telemetry.

When the trigger fires, this field is resurfaced automatically. Watchable rather than parked.

depends onROI

Trigger · A Nightingale measurement run on a client engagement shows fewer hours billed for the same output — substitution, not augmentation — with the client's telemetry, not a survey.

When the trigger fires, this field is resurfaced automatically. Watchable rather than parked.

enablingCost redux on tokens

The price curve is the premise; sa-cost-per-mtok is the number this field runs on.

compoundingROI

A Nightingale ROI measurement is the only way to distinguish substitution from augmentation.

compoundingDark Factory

Dark functions are where labour pricing actually changes; this field is the abstract end of that one.

blockingDeciding Table Stakes

If analysis is table stakes, the firm cannot price it at all; the field's consequence depends on that one's verdict.

Share graph

Provenance running forward.

Discovery, not accountability. No counts, no rankings, no rollups to managers.

Convergence · who else is here

Several people’s drops meet here. An informal working group already exists and probably does not know it.

ContributorsAWRMJPCD

Lineage

What this field produced, and what it killed.

Experiments, recommendations and graveyard entries stay attached. The reasoning that killed a claim is the reusable asset.

No experiments, recommendations or graveyard entries yet. That is what a candidate looks like.

Open questions · return to the pile

Every run leaves a record. Separately, its question either closes or returns to the pile with notes — which is what the next person proposing the same thing will see.

  1. 01Is the substitution figure a leading indicator for senior hours or a permanent junior-only effect?
  2. 02At what token cost per task does a client stop asking for a day rate, and has that already happened in banking?
  3. 03Should this be elected at one horizon and falsified, or is the two-horizon placement doing honest work?