Analytic Standards & Methodology

How ODIN produces intelligence — and why every product it generates can be audited.

ODIN is built on the same analytic tradecraft standards used by professional intelligence organizations. The difference is enforcement. Where traditional operations rely on training and human discipline, ODIN applies these standards systematically — on every report, every assessment, every export. Nothing is graded by mood, and nothing leaves the system without a source trail. This page explains each standard we enforce and what it means for the intelligence you receive.

Every ODIN product opens with the conclusion. Assessments lead with a direct answer — the bottom line — followed by key judgments and then the supporting analysis. You are never asked to read four pages to find out what we think.

This is the standard structure of professional intelligence writing, and it exists for one reason: decision-makers act under time pressure. The analysis is always there for those who want to verify the reasoning. The answer comes first.

Every source ODIN draws on carries a two-part grade: a letter for the reliability of the source itself, and a number for the credibility of the specific information. A grade of B2 means a usually-reliable source providing information that is probably true. These grades appear as badges alongside citations in ODIN's responses and in the Source Log appendix of every exported brief.

Source Reliability

Information Credibility

ODIN never says "might," "could," or "possibly" and leaves you to guess what that means. When ODIN expresses likelihood, it uses only the calibrated terms defined in Intelligence Community Directive 203 — each mapped to a defined probability range:

Alongside likelihood, ODIN states its analytic confidence — HIGH, MODERATE, or LOW — based on the quality and corroboration of the underlying sources. The two are never blended. "Likely" tells you the odds of the event; confidence tells you how solid the sourcing behind that estimate is.

When a field report enters ODIN, it is evaluated against the client's own Standard Operating Procedures — typically in under two minutes. Every reported action is graded in a three-column tactical matrix:

Each action receives one of three severity ratings:

• COMPLIANT — action met or exceeded SOP requirements
• DEVIATION — action departed from SOP; corrective guidance issued
• CRITICAL VIOLATION — action created unacceptable risk or breached a mandatory protocol

Grading is performed against the written SOP, not against opinion. The same report, evaluated twice, produces the same result — which is the point.

Severity ratings drive action automatically. When ODIN identifies a Critical Violation, its escalation engine fires immediately: dual-channel system alerts and direct supervisor notification, generated and dispatched without waiting for a human to read the report first. Deviations are logged and surfaced for supervisor review; Critical Violations interrupt.

The escalation thresholds are doctrine, not discretion. What triggers an alert is defined in advance with the client — so at 3 AM on a holiday weekend, the standard is exactly what it was at 10 AM on a Tuesday.

Every ODIN product is traceable. Each response and exported brief carries a unique report reference ID (format: ODIN-YYYYMMDD-XXXX) and a Source Log recording:

• ▸ Every source consulted, with its Admiralty grade and grading rationale
• ▸ Timestamp of the query (UTC)
• ▸ The system version that produced the assessment

Exported intelligence briefs include the full Source Log as a chain-of-custody appendix. If an assessment is ever questioned — by leadership, by counsel, by an auditor — the complete evidentiary basis is attached to the document itself.

ODIN's assessments draw on a continuously maintained doctrine library — spanning tactical, regulatory, and situational doctrine — alongside live operational feeds, including National Weather Service data and municipal CAD activity where available. Field reports from client sites enter the same pipeline. Every input is graded on ingestion; nothing enters the analytic layer ungraded.

ODIN's standards are enforced by the system, and they cut both ways. We do not overstate confidence to sound decisive. We do not bury conclusions to sound thorough. We do not cite what we cannot grade, and we do not grade what we cannot trace. When ODIN doesn't know, it tells you — in those words.

That is the standard. Every product, every time.