How is sensitive data protected?

All documents are encrypted with AES-256-GCM at rest and TLS 1.3 in transit. A GDPR anonymization layer removes PII before any AI processing. Your data never leaves our SOC 2 certified infrastructure.

How long does integration take?

Less than 30 minutes. Upload documents directly, connect via OAuth for Slack, or use our REST API for bulk processing.

How does outcome tracking work?

Outcomes are detected automatically from follow-up documents, Slack messages, and web intelligence. You confirm with one click. Each reported outcome makes your future analyses more accurate.

How is Decision Intel different from ChatGPT?

ChatGPT gives one opinion from one model. Decision Intel uses 3 independent judges for noise measurement, a bias interaction matrix for compound scoring, 30+ cognitive biases benchmarked across 135 historical decisions, and an outcome flywheel that gets smarter with every confirmed decision.

Live demo: Decision Intel on Long-Term Capital Management (1998)

Top-line result

F28/100

Decision Quality Index · Critical

The presence of two Nobel laureates made the model un-challengeable internally (authority bias), external risk voices were absent (no non-principal risk committee), and the 1997 capital return increased rather than decreased effective leverage — a decision only explicable under overconfidence.

Overconfidence BiasAuthority BiasGroupthink

Uploaded Document

The excerpt the platform analyzed

“1997 annual investor letter describing 'negligible' portfolio risk based on Value-at-Risk models”

Source:Long-Term Capital Management Partners (John Meriwether)·1997

Platform-detected signals

4 red flags in the document

Each flag below was detectable from the document text alone. No outcome data. No hindsight.

01

VaR models assuming normal distributions in fat-tailed markets

Critical · Detectable at decision time

02

Leverage exceeding 25:1 on notional $1.25T

Critical · Detectable at decision time

03

Correlation assumptions based on 5-year lookback missing regime changes

Critical · Detectable at decision time

04

Nobel laureate credentials creating authority bias in risk oversight

Critical · Detectable at decision time

Cognitive biases detected

6 biases across the document

★ Overconfidence BiasAnchoring BiasCognitive MiseringGroupthinkConfirmation BiasGambler's Fallacy

CEO / Board Question Simulator

Questions the steering committee would ask

Predicted from the document's own signals. In the live product, these are generated from your memo — no two runs produce the same list.

“Is there a chance we're underestimating overconfidence bias risk on this decision — and what would change our mind if so?”

“Walk me through vaR models assuming normal distributions in fat-tailed markets — what's our defensible answer if the board raises it?”

“What would it cost us to do retain 1997 capital base (avoid the forced leverage increase); cap gross leverage at 10:1; diversify across strategies with genuinely uncorrelated drivers; i… before we commit?”

“If this memo were re-run in six months with the outcome revealed, what do we think would embarrass us most?”

Counterfactual

What a bias-adjusted process would have done

Recommendation

Retain 1997 capital base (avoid the forced leverage increase); cap gross leverage at 10:1; diversify across strategies with genuinely uncorrelated drivers; install an independent non-principal risk officer with veto power.

Why it would have worked

Every bias LTCM exhibited traces to a closed-loop where the same people who designed the models also signed off on leverage and rejected external challenge. The model wasn't the problem — absence of any external risk voice was.

Estimated alternative outcome

Fund survives Russia default with 15–20% drawdown rather than 44%; likely continues operating through the early 2000s before gradual wind-down.

Evidence Timeline

What was visible, and when

Every event below was documentable before the outcome was known. The platform looks for signals like these in live memos.

1994-02
LTCM launches with $1.25B from 80 investors at 25× leverage; initial principals include Meriwether, Scholes, Merton, and ex-Fed Vice Chairman David Mullins.
PWG Report, Appendix A
1996
Returns of 41% net of fees; fund becomes unable to deploy new capital at same convergence spreads, signaling trade crowding.
Lowenstein, "When Genius Failed", ch. 7
1997-12
Fund returns $2.7B to investors — but retains full trading book, actually *increasing* leverage relative to equity base.
LTCM 1997 investor letter
1998-05
Salomon Smith Barney closes its arbitrage desk, unwinding positions similar to LTCM — the first warning that "convergence" strategies were becoming divergent.
PWG Report, ch. 3
1998-08-17
Russia defaults on ruble-denominated debt; flight-to-quality drives spreads wider rather than narrower. LTCM loses 44% of capital in one month.
Federal Reserve Board staff study (Kambhu et al., 2007)
1998-09-23
Fed-organized consortium of 14 banks injects $3.625B to prevent forced liquidation; LTCM effectively dissolved.
PWG Report, ch. 4

Who Was in the Room

Stakeholders and positions

Who advocated, who dissented, who was overruled, and who stayed silent — the most reliable single signal of decision-process quality.

John Meriwether

Advocate

Founding Principal & CEO

Pushed to return capital in 1997 to keep per-partner returns high rather than shrink leverage.

Myron Scholes

Advocate

Principal (Nobel Laureate, Options Pricing)

Publicly defended the model framework; later acknowledged leverage as the core error.

Robert Merton

Advocate

Principal (Nobel Laureate, Continuous-Time Finance)

David Mullins

Silent

Principal (former Fed Vice Chair)

Brought regulatory gravitas but did not publicly flag leverage or crowding concerns.

Eric Rosenfeld

Advocate

Principal (risk & trading systems)

Cross-Case Pattern

Other decisions with the same pattern

Your memo, not Long-Term Capital Management's

Run your own strategic memo in 60 seconds.

Same bias framework, same pre-decision evidence discipline. Upload a strategic memo or board deck — get the DQI grade, flagged biases, predicted questions, and counterfactual back before your next meeting.

Start free audit See more demos

4 red flags in the document

Each flag below was detectable from the document text alone. No outcome data. No hindsight.

01

VaR models assuming normal distributions in fat-tailed markets

Critical · Detectable at decision time

02

Leverage exceeding 25:1 on notional $1.25T

Critical · Detectable at decision time

03

Correlation assumptions based on 5-year lookback missing regime changes

Critical · Detectable at decision time

04

Nobel laureate credentials creating authority bias in risk oversight

Critical · Detectable at decision time

Questions the steering committee would ask

Predicted from the document's own signals. In the live product, these are generated from your memo — no two runs produce the same list.

“Is there a chance we're underestimating overconfidence bias risk on this decision — and what would change our mind if so?”

“Walk me through vaR models assuming normal distributions in fat-tailed markets — what's our defensible answer if the board raises it?”

“If this memo were re-run in six months with the outcome revealed, what do we think would embarrass us most?”

What a bias-adjusted process would have done

Recommendation

Why it would have worked

Estimated alternative outcome

Fund survives Russia default with 15–20% drawdown rather than 44%; likely continues operating through the early 2000s before gradual wind-down.

What was visible, and when

Every event below was documentable before the outcome was known. The platform looks for signals like these in live memos.

1994-02

LTCM launches with $1.25B from 80 investors at 25× leverage; initial principals include Meriwether, Scholes, Merton, and ex-Fed Vice Chairman David Mullins.

PWG Report, Appendix A

1996

Returns of 41% net of fees; fund becomes unable to deploy new capital at same convergence spreads, signaling trade crowding.

Lowenstein, "When Genius Failed", ch. 7

1997-12

Fund returns $2.7B to investors — but retains full trading book, actually *increasing* leverage relative to equity base.

LTCM 1997 investor letter

1998-05

Salomon Smith Barney closes its arbitrage desk, unwinding positions similar to LTCM — the first warning that "convergence" strategies were becoming divergent.

PWG Report, ch. 3

1998-08-17

Russia defaults on ruble-denominated debt; flight-to-quality drives spreads wider rather than narrower. LTCM loses 44% of capital in one month.

Federal Reserve Board staff study (Kambhu et al., 2007)

1998-09-23

Fed-organized consortium of 14 banks injects $3.625B to prevent forced liquidation; LTCM effectively dissolved.

PWG Report, ch. 4

Stakeholders and positions

Who advocated, who dissented, who was overruled, and who stayed silent — the most reliable single signal of decision-process quality.

John Meriwether

Advocate

Founding Principal & CEO

Pushed to return capital in 1997 to keep per-partner returns high rather than shrink leverage.

Myron Scholes

Advocate

Principal (Nobel Laureate, Options Pricing)

Publicly defended the model framework; later acknowledged leverage as the core error.

Robert Merton

Advocate

Principal (Nobel Laureate, Continuous-Time Finance)

David Mullins

Silent

Principal (former Fed Vice Chair)

Brought regulatory gravitas but did not publicly flag leverage or crowding concerns.

Eric Rosenfeld

Advocate

Principal (risk & trading systems)