Every question a CSO, GC, or vendor-risk reviewer asks before signing.
Category, methodology, security, procurement, answered in one place with the canonical sources one click away. Updated whenever the methodology version, framework registry, or taxonomy ships an extension.
Category + product
What Decision Intel is, and who it is for
The category claim, the buyer types, and the contrast against the tooling already on the procurement reader’s shortlist.
What is Decision Intel?
Decision Intel is the reasoning audit platform. Most tools audit your data. We audit your reasoning — and catch the fatal blind spots in strategic memos before the committee does. Chief Strategy Officers, corporate development teams, fund partners, and PE-backed CEOs run every strategic memo, board deck, and IC artefact through the audit before the committee sees it — and the reasoning trail compounds quarter after quarter into a living Decision Knowledge Graph that audit committees can defend.
It is the protected category noun. BI tools audit your data. Model-risk-management tools audit your algorithms. the reasoning audit platform audits the human reasoning chain that produced the recommendation. The 22-bias canonical taxonomy fires on the memo text, the 22×22 bias-interaction matrix surfaces compound failure patterns, and the procurement-grade Decision Provenance Record carries the audit trail into the legal record.
Four ICP personas: fractional Chief Strategy Officers running 3-5 client engagements, Heads of Corporate Development at scale-ups paying personally pre-team-budget, partners at smaller funds with active deal flow or investor-governance pressure, and PE-backed mid-market founders. The Fortune 500 corporate-strategy ceiling activates after the wedge: cross-border M&A leaders whose audit committees need the regulatory-mapping moat.
Capital eroded by unaudited reasoning in strategic decisions. Reasoning is never objectively sound; it is either audited or unaudited. Capital is not destroyed because executives have cognitive biases; biases are the operating system of the human mind. Capital is destroyed because organisations lack the structural friction required to catch and neutralise bias before capital is committed.
How is this different from ChatGPT or a general AI assistant?
ChatGPT gives one opinion from one model: ungoverned, untraceable, unaudited. the reasoning audit platform runs a 12-node analysis pipeline with a 3-frame noise jury (analyst-skeptical, regulator-hostile, contrarian-strategist) across two model families, arbitrated by a metaJudge, scored against a 143-case reference library, and persisted as a hashed tamper-evident artefact. Not a chatbot; a reasoning audit, checkable from memo to outcome.
Cloverpop logs decisions; Decision Intel audits them. Cloverpop is positioned as a decision system of record (logging + voting + accountability) but has no bias detection. The pain framing "unaudited reasoning" is precisely the gap they cannot close without our 22-bias taxonomy + R²F + 143-case corpus.
How is Decision Intel different from IBM watsonx.governance?
IBM audits the model; Decision Intel audits the human reasoning. IBM watsonx audits AI model behaviour (model lineage, fairness metrics, drift detection). It does NOT audit the human-authored memo or the human reasoning chain that drove the capital-allocation decision. The "human reasoning" qualifier is the procurement-grade differentiator.
Red teams fail structurally because the political cost of dissent in sponsor-driven environments is unsustainable. The antagonist that costs you no political capital fires before the IC memo can hide what the deal sponsor does not want to see. The audit surfaces the dissent algorithmically, so the corp-dev professional shifts from antagonist trying to kill the sponsor’s deal to facilitator surfacing a system-generated risk flag. Same dissent, zero ego cost.
Methodology + IP
How the audit works, and the academic anchors behind it
The Recognition-Rigor Framework, the bias taxonomy, the Decision Provenance Record, the Decision Quality Index. Every claim traceable to a primary source.
What is the Recognition-Rigor Framework (R²F)?
R²F is the protected IP moat. Kahneman’s System 2 debiasing pipeline (bias detection, noise jury, statistical scoring) and Klein’s Recognition-Primed Decision framework (pattern recognition, mental simulation, pre-mortem) arbitrated in one pipeline by a metaJudge. The only vendor running both halves. Anchored on Kahneman & Klein (2009) "Conditions for Intuitive Expertise: A Failure to Disagree."
How many cognitive biases does the platform detect?
22 biases in the canonical taxonomy, stable IDs DI-B-001 through DI-B-022. Each carries a real-world example, debiasing techniques, related biases, and a primary academic citation with DOI. The narrowness is the moat. Every detector is paper-grounded, not heuristically added.
Every audit produces a Decision Provenance Record: hashed, tamper-evident, with SHA-256 input hashes, methodology version stamp (current: 2.4.0), prompt fingerprint, DQI weight-resolution hash, and a composed Evidentiary Standard fingerprint bound into the legal trail. Mapped onto EU AI Act Article 14, Basel III Pillar 2 ICAAP, SEC AI disclosure, GDPR Article 22, and the 11 AI Verify Foundation principles.
A weighted composite score derived from seven components (bias load, noise, evidence quality, process maturity, compliance exposure, historical alignment, compound risk) calibrated against 143 historical corporate decisions in the public reference library. Methodology version 2.4.0. Weights are user-adjustable on the Strategy tier per the Dietvorst (2015) algorithm-aversion fix: practitioners use imperfect algorithms IF allowed to slightly modify the inputs or weights.
The first public ranking of which biases predict failure by industry, built from the case library and (as customers consent) calibrated against live outcome data. Every metric carries its sample size; dimmed rows flag n<3. The cross-org data flywheel that compounds the platform’s defensive moat against incumbents who lack reasoning-quality data.
A living record of every strategic decision your team has run through the platform, their outcomes, and the reasoning trail. Decision history survives team transitions (CSO leaves; reasoning trail stays). Audit-committee Q&A pulls up the reasoning in 60 seconds. Future decisions get sharper because the platform learns YOUR specific bias patterns via Brier-scored per-org recalibration.
What academic research underpins the platform?
Kahneman & Klein (2009) "Conditions for Intuitive Expertise" is the canonical anchor for the Recognition-Rigor Framework. Kahneman & Lovallo (2003) "Delusions of Success" grounds reference-class forecasting. Klein & Mitchell (1995) and Mitchell, Russo & Pennington (1989) ground the prospective-hindsight pre-mortem. Dawes (1979) "The Robust Beauty of Improper Linear Models" grounds the decision-rubric detector. Dietvorst, Simmons & Massey (2015) grounds the algorithm-aversion counter-programming. Every academic anchor carries its DOI in the bias-education taxonomy.
Every closed outcome is Brier-scored against the predicted Decision Quality Index. Per-org calibration data sharpens the model for that organisation’s specific bias patterns. The Outcome Gate (enforced on HXC-cohort accounts from day one) ensures the loop closes: a user cannot escalate to the next decision until the prior decision’s outcome is on record. Engineered into the workflow, not optional.
Security + procurement
Posture for vendor-risk registers and audit committees
Encryption, regulatory mapping, sub-processors, retention SLA, indemnification, and the contracted artefacts a procurement reviewer needs to clear a Wednesday queue.
What is the security posture?
Hosted on SOC 2 Type II infrastructure (Vercel + Supabase). Decision Intel’s own product-level SOC 2 Type I audit is targeted for Q4 2026, with the Type II observation window opening immediately after; in-flight controls already mirror Type II.
Which regulatory frameworks does the DPR map onto?
19 frameworks across G7, EU, GCC, and African markets. The anchor set covers EU AI Act Article 14 (human oversight), Basel III Pillar 2 ICAAP (qualitative-decision documentation), SOX §404 (internal controls), SEC AI disclosure, GDPR Article 22 (automated-decision rights), FCA Consumer Duty, NDPR (Nigeria), CBN AI guidance, WAEMU, CMA Kenya, POPIA (South Africa), SARB Model Risk, and seven more. Every bias finding in a DPR carries its regulatory exposure inline.
Where is data stored, and who are the sub-processors?
Production data lives in US-region Supabase (Postgres + Auth) with AES-256-GCM at rest and TLS 1.2+ in transit. Application compute runs on Vercel. Email delivery via Resend. DNS + edge via Cloudflare. AI inference via Anthropic (Claude) and Google (Gemini) under no-training contractual terms. The full sub-processor schedule with verification paths is on the trust page.
Three tiers. Individual: 1 year (legal-defensible floor). Strategy (team): 3 years (mid-market default for quarterly board cycles). Enterprise: 7 years (SOX §404 aligned). Custom retention (HIPAA, banking, government) negotiable on pilot agreement. Every entry is immutable, append-only, timestamped at write, queryable via the AdminAuditLog UI, and exportable as a JSON bundle on Enterprise.
What is the indemnification posture?
Standard 12-months-of-fees cap with carve-outs for confidentiality breaches, wilful misconduct, third-party IP claims, and sub-processor failures. Cyber-liability + E&O insurance procurement scheduled for Q1 2027. Posture surfaced on /security and contracted in the DPA template available for download.
No. Customer content is not used to train models, contractually locked in the DPA and reinforced via the vendor agreement with every AI sub-processor (Anthropic and Google both honour no-training enterprise contracts). The cross-org Bias Genome aggregates outcome metadata only, never raw content, and only across consenting accounts.
Hosted on SOC 2 Type II infrastructure (Vercel + Supabase, both Type II audited). Decision Intel’s own SOC 2 Type I is targeted for Q4 2026 with Type II observation window opening immediately after. The DPA, sub-processor schedule, vendor-risk questionnaire, and DPR specimens are all available on the trust page in advance.
Who founded Decision Intel, and what stage is the company at?
Decision Intel was founded in 2024 by Folahan Williams, Founder. The company is in design-partner phase: a five-seat program for Fortune 500 strategy teams shaping the Recognition-Rigor Framework as it scales toward enterprise general availability. Legal entity, jurisdiction, registered office, and procurement contact are disclosed on the About page.
Procurement, security, audit-committee, and compliance questions go to the founder directly. 24-hour acknowledgement, 5-business-day substantive response.