Crene is investment thesis review infrastructure. The platform decomposes macro, labor, AI transition, and investment thesis questions into falsifiable components using four frontier AI models (Claude, GPT, Gemini, Grok). The system produces three product types: thesis maps (binary decomposition into assumptions), factors (continuous decomposition into drivers), and scenarios (hybrid decomposition into coherent pathways). Cross model disagreement is the primary signal. Calibrated against 1,161 resolved events with a Brier score of 0.114.

How does AI prediction calibration work at Crene?

Calibration applies at the consensus layer, not at the underlying model layer. Crene queries four frontier models in isolation, computes a cross model median consensus, and tracks resolution against a tiered source allowlist. Every resolved event becomes a permanent calibration point. Per tier accuracy, per model Brier scores, and per domain accuracy are computed live and published at https://crene.com/methodology.

What is a Brier score?

A Brier score is the proper scoring rule for probability forecasts on binary outcomes. It is the mean squared error between the forecast probability and the realized outcome (0 or 1). Lower scores indicate better calibration. The no skill baseline is 0.25 (always predicting 0.5); a perfect forecaster scores 0.0. Crene publishes the consensus Brier score and per model Brier scores across the resolved event corpus.

Crene serves investment teams that need a repeatable thesis review process: hedge funds, asset managers, macro PMs, CIO offices, family offices, and thematic investment teams. A team brings one thesis it is actively debating. Crene maps what the thesis depends on, where models disagree, what changed each week, and what would force a rethink. Data and API access remain available as the proof layer behind the workflow.

What is the difference between an event, a thesis map, and a factor at Crene?

Crene has four product types. Events are base binary outcomes with scalar probabilities. Thesis maps decompose binary anchor questions into child situations. Factor maps decompose continuous variables into drivers. Scenarios map long horizon investment questions into coherent pathways across live systems including Empire by Default, AI Labor Transition, European Rearmament, and India as the Third Growth Pole.

The assumption layer underneath investment decisions

Map the assumptions behind one live thesis.

Crene turns a live market view into an inspectable map: what it depends on, where models disagree, what changed this week, and what would force the team to rethink.

View examples Review a thesis

AI displacementDollar dominanceDemographic collapseFiscal stress

One thesis, decomposed

AI Economic Realization

4-model ensemble

Strategic question

Will AI infrastructure convert into measurable productivity, earnings, and GDP impact?

Productivity transmission

Hinge family

Workflow adoption, output per worker, process redesign

Market realization

Model disagreement

Revenue lift, margin expansion, capex efficiency

Measurement lag

Update conditions

GDP attribution, earnings evidence, adoption depth

Output

Assumptions that matterWhere models disagreeWhat would change the view

Example thesis map

AI Economic Realization, mapped into assumptions.

The map starts with a live AI thesis and shows the assumptions underneath it: enterprise adoption, productivity pass through, capex efficiency, labor substitution, margin evidence, and GDP measurement.

Open full ontology →

Ontology index

The assumption layer, compressed for mobile.

The desktop map shows the full constellation. On mobile, Crene shows the same structure as an index: public anchors, shared ontology fields, and the live nodes attached to each map.

10anchors

12fields

578nodes

Shared ontology fields

MarketsOtherAI laborAI productivityFiscal capacityRates pathGrowth polesDollar systemAI contentDemographicsInflationDefense capacity

Live anchors

ScenarioIndia as the Third Growth Pole

29 nodes

ScenarioEuropean Rearmament

31 nodes

ScenarioAI Labor Transition

168 nodes

ScenarioEmpire by Default

38 nodes

Thesis mapWarsh Fed Rate Path

100 nodes

Thesis mapAI Economic Realization

109 nodes

Thesis mapAI Content Dominance

100 nodes

Factor mapUS GDP Directly Attributable to AI Systems, 2030

1 nodes

Factor mapS&P 500 close, year-end 2026

1 nodes

Factor mapUS 10Y Treasury Yield, year-end 2026

1 nodes

Open this page on desktop to inspect the full constellation layout.

449 visible of 578 nodes10 shared ontology threadsClick cores, fields, or nodes to inspect.

Inspector

Crene Knowledge Map

Combined ontology constellation

Shared assumption threads10 total

Fields that recur across separate live maps. These are the assumptions that can move more than one thesis.

AI Economic Realization

One AI thesis, decomposed.

Open the full map to inspect how AI adoption, productivity, capex, margins, labor, and GDP measurement connect across the thesis.

Open full ontology →

Guided live example

See the public Warsh/Fed map, then bring your own thesis.

The live map is proof of the method. The pilot is the product workflow. Use the public example to see how Crene structures one debated macro thesis, then submit a thesis your own team needs to review.

View Warsh/Fed example Map one thesis

Start with a live thesis

Warsh/Fed rate path is a real macro question with a fixed horizon and live model disagreement.

Inspect what carries it

Crene shows the assumptions, pressure points, and places where the models do not agree.

Bring your own thesis

After seeing the public example, submit one question your team is actively debating.

This week in your thesis

The weekly review surface, not another dashboard.

Crene should make a team feel the Monday meeting rhythm: what moved, where the thesis is fragile, and what would force a rethink before the next IC, risk review, or strategy meeting.

Consensus moved

Base case moved from 45% to 52% after the policy path tightened.

Largest disagreement

Models now disagree most on labor resilience, not AI capex.

Rethink threshold

Two assumptions crossed review thresholds and need owner review.

New evidence

One new source changed the read on the policy reaction function.

View private thesis demo Map one thesis

Why this compounds

Every thesis map becomes institutional memory.

A single map clarifies the current decision. A sequence of maps records how the institution updates judgment over time.

What the team believed

Which assumptions carried the view

What changed this week

Which assumptions were wrong

Which drivers recur across theses

Which rethink triggers fired

Thesis review workflow

Bring a thesis. Get a weekly review surface.

Start with one live market view. Crene turns it into assumptions, disagreement, rethink triggers, and weekly changes your team can review before the next meeting.

01

Bring one thesis

Send a live view your team is already debating.

02

Map the assumptions

Break the thesis into drivers, dependencies, and observable checks.

03

Find disagreement

See where model views split and which assumptions carry the debate.

04

Review what changed

Track new evidence, pressure points, and rethink triggers each week.

Live assumption maps

Active investment theses.

Scenario maps track long-horizon uncertainty through structured pathways, assumption families, and observable update conditions.

All examples →

Scenario map

38 tracked

Empire by Default

Will the United States account for more than 30% of global nominal GDP by 2050?

55 accelerating35 resisting10 mixed

Scenario map

168 tracked

AI Labor Transition

What determines whether AI expands, reduces, or restructures productive work in the United States by 2030?

30 accelerating35 resisting35 mixed

Scenario map

29 tracked

India as the Third Growth Pole

Does India become a durable third pole of global growth, capital formation, and technology demand by 2035?

47 accelerating35 resisting18 mixed

Scenario map

31 tracked

European Rearmament

Can Europe convert threat perception into durable fiscal, industrial, and military capacity by 2030?

46 accelerating36 resisting18 mixed

Calibration record

Resolved scoring is reported separately from still-accruing product maps.

Macro Brier0.2304n=262, ex earnings

External benchmark0.114n=811

Product layer10 mapslive product layer

View methodology →

Thesis maps

Track the structure under a thesis.

Binary questions broken into observable components, model consensus, disagreement, and resolution criteria.

All thesis maps →

FED POLICY

100 components

Warsh Fed Rate Path

Whether the Warsh Fed converts its hawkish June pivot into an actual 2026 rate hike. 100 child conditions spanning policy path, Fed communications, inflation, labor resilience, institutional reset, market pricing, and consensus narrative.

47% consensus100 components

AI ADOPTION

109 components

AI Economic Realization 2026

Whether realized enterprise AI absorption lags infrastructure investment. Nine weighted settlement indicators across six causal domains, plus 100 supporting children.

70% consensus109 components

AI CONTENT DOMINANCE

100 components

AI Content Dominance by 2030

Will AI generated content exceed human generated content on the US internet before 2030? 100 child situations spanning platform dynamics, search and SEO, journalism, video and visual content, academic integrity, authentication infrastructure, economic disruption, legal frameworks, societal trust, and compute infrastructure.

72% consensus100 components

Factor maps

Follow variables as distributions.

Numerical variables decomposed into drivers, percentile distributions, model spread, and update trajectories.

All factor maps →

ai economics

Factor map

US GDP Directly Attributable to AI Systems, 2030

What percentage of US GDP will be directly attributable to AI systems by 2030?

6.10% p50±2.22% spreadLOW

equity index

Factor map

S&P 500 close, year-end 2026

What will the S&P 500 Index (SPX) official close be on the final trading day of 2026?

7,742 p50±259 spreadMEDIUM

rates

Factor map

US 10Y Treasury Yield, year-end 2026

What will the US 10-year Treasury constant maturity yield be at the last Treasury market close of 2026?

4.03% p50±0.37% spreadHIGH

Disciplined use

Built for structured judgment, not black box signals.

Crene is designed for teams that need inspectable assumptions, model disagreement, and update conditions before evidence is complete.

Not market predictions

Crene does not claim to forecast traded prices or replace market signals.

Not investment advice

Outputs are research infrastructure for internal analysis and decision support.

Not a black box signal

Every map is organized around assumptions, disagreement, and what would change the view.

Access

Bring one live market thesis.

Crene maps one investment thesis your team is already debating, then tracks the assumptions, model disagreement, and update conditions over time.

Review a thesis

API

Build with live endpoints.

Public read endpoints for live and resolved Crene objects. JSON responses, stable URLs, and private write access for pilot customers.

API documentation

Built by Stephen Lee. Seven years in institutional finance at Goldman Sachs and Credit Agricole, now building Crene for investment teams reviewing strategic theses.

About →