Methodology Overview - Alpha Documentation

Alpha Governance Ratings provide independent, evidence-based assessments of how companies govern artificial intelligence. The rating system functions as a credit rating analog for AI governance, designed for board directors, institutional investors, C-suite executives, and governance professionals making fiduciary decisions.

The Question We Answer

Is this company governing AI in a way that protects stakeholders and enables sustainable competitive advantage? We measure governance effectiveness, not AI capability or strategy quality. We measure outcomes and stress responses, not just governance artifacts. A published policy earns partial credit. Evidence that the policy was applied in a real decision earns full credit. Evidence that governance held under competitive pressure earns the highest marks.

The 4-Phase Alpha Standard

Every rating follows a structured, repeatable process:

Phase	Name	Description
1	Evidence Collection	Gather public filings (10-K, DEF 14A, proxy statements), corporate disclosures, safety reports, news coverage, and proprietary signals from the Alpha Governance Knowledge Graph
2	Dimension Scoring	Score each of the 12 dimensions (D1-D12) using codified indicators, weighted by the company’s Rating Profile
3	Cross-Dimension Analysis	Apply concern ceiling logic, critical floor checks, cross-signal inference, and committee zone coherence analysis
4	Grade Assignment	Calculate composite score, assign letter grade (AAA-D), determine trajectory and watch status

Two-Track Rating Architecture

Alpha operates a two-track system to balance independence with depth:

Public Profile

Available for all covered companies. No letter grade. Full analytical content including materiality profile, dimension analysis, signal flags, peer comparison, and regulatory exposure mapping. Based on publicly available evidence.

Assessed Rating

Full AAA-D letter grade. Available through direct engagement. Includes proprietary signal analysis, founder review by Steven Wolfe Pereira, and remediation roadmap. Required for enterprise compliance workflows.

Rating Profiles

Different companies face different governance challenges. Alpha uses three profiles within the same dimensional framework:

Profile	Description	Higher Weight On
Developer	Companies that build, train, and distribute AI models	Safety (D2), Data Ethics (D4), Cybersecurity (D9)
Deployer	Companies that integrate, deploy, and operate AI systems	Governance Infrastructure (D5), Fairness (D6), Workforce Impact (D7), Supply Chain (D10)
Hybrid	Companies that both build and deploy	Blended weights across all dimensions

The rating report always discloses which profile was applied and why.

Evidence Tiers

All evidence is classified by reliability:

Tier	Source Type	Weight Multiplier
T1	Regulatory actions, enforcement, court filings	3x
T2	Verified internal documents, audit reports	3x
T3	Expert assessments, structured interviews	3x
T4	Proprietary signal analysis (AGKG)	3x
T5	Public disclosures, news, corporate websites	1x

A rating that relies solely on T5 signals is flagged with data tier limitations in the output.

Independence

Alpha rates companies on public data without seeking permission or offering courtesy reviews. We do not partner with rated entities. Independence creates credibility. When a company has an active paid advisory engagement, the rating requires direct founder review by Steven Wolfe Pereira with conflict disclosure on record.

What’s Next

12 Dimensions

The full D1-D12 framework with indicators and scoring criteria

Scoring Rubric

How Alpha converts evidence into scores, confidence, floors, and final grades

Rating Scale

AAA to D: what each grade means

Committee Mapping

How dimensions map to board committee accountability

Integrity Screen

Anti-gaming protections and disclosure timing flags

​The Question We Answer

​The 4-Phase Alpha Standard

​Two-Track Rating Architecture