Diagnosis Engine

How Diagnosis Works

The diagnosis engine takes mapped workflow data from the observation phase and runs it through a multi-layer analysis pipeline. Each layer examines a different aspect of your operations: individual workflow efficiency, cross-team collaboration patterns, and system-level bottlenecks.

Analysis Layers

Layer 1: Workflow Efficiency

Examines individual workflow types (e.g., bug fix lifecycle, feature request pipeline) and measures actual cycle time against statistical baselines from comparable organizations.

• Identifies workflows with abnormally high cycle times
• Detects excessive handoff points that add latency
• Flags rework loops where tasks repeatedly move backward in the pipeline

Layer 2: Collaboration Patterns

Analyzes cross-tool communication patterns to understand how teams coordinate. Identifies bottleneck individuals (people who appear as blockers in multiple workflow types) and communication gaps.

• Communication density analysis across team boundaries
• Blocking dependency detection in approval chains
• Information silo identification (teams that should communicate but don't)

Layer 3: Pattern Matching

Compares your organization's patterns against our database of 200+ documented operational patterns. Each match includes a confidence score, impact estimate, and recommended intervention type.

Configuration

Confidence Threshold

Controls the minimum confidence required for a pattern match to be included in recommendations. Lower thresholds surface more potential opportunities but may include less certain findings.

Industry Focus

Bias the pattern matching toward industry-specific patterns. This narrows the pattern database to the most relevant comparisons.

{
  "type": "diagnosis",
  "config": {
    "confidenceThreshold": "standard",
    "industryFocus": "software_development",
    "excludePatterns": [],
    "weightOverrides": {
      "impact": 1.0,
      "complexity": 1.0,
      "readiness": 1.2,
      "dataAvailability": 0.8
    }
  }
}

Score Weight Overrides

Customize how the four scoring dimensions are weighted in the final opportunity ranking. Increase the weight of dimensions that matter most to your organization:

• impact: Prioritize highest-impact opportunities
• complexity: Penalize complex implementations more heavily
• readiness: Prioritize opportunities where teams are most ready
• dataAvailability: Weight data quality considerations

Output Format

The diagnosis engine produces a ranked list of AI opportunities, each including:

• Composite score (0–100) based on weighted dimensions
• Matched pattern reference with confidence percentage
• Estimated time savings per week
• Implementation complexity rating (low / medium / high)
• Recommended agent type and deployment approach

Previous: Observation Agents | Next: Deployment