CLAGA Adaptation Flow

Overview

This diagram illustrates how Cognitive Load Aware Growth Agents (CLAGAs) detect human cognitive state in real-time and adapt information delivery to prevent overwhelm and ensure actionable intelligence.

Diagram

graph TB
    subgraph "User Interaction Layer"
        UserAction[User Actions<br/>Clicks, Pace, Queries]
        ContextSignals[Context Signals<br/>Time, Day, Calendar]
        HistoricalData[Historical Patterns<br/>Past Behavior, Stress Markers]
    end
    
    subgraph "CLAGA Detection Engine"
        SignalCollection[Signal Collection<br/>---<br/>Interaction Speed<br/>Query Complexity<br/>Navigation Patterns]
        
        StateAnalysis[Cognitive State Analysis<br/>---<br/>Load Level Detection<br/>Stress Indicators<br/>Capacity Assessment]
        
        UserAction --> SignalCollection
        ContextSignals --> SignalCollection
        HistoricalData --> SignalCollection
        
        SignalCollection --> StateAnalysis
    end
    
    subgraph "Load State Classification"
        StateAnalysis --> LoadDecision{Cognitive<br/>Load Level?}
        
        LoadDecision -->|Low Load<br/>Calm, Exploratory| LowLoad[LOW LOAD STATE<br/>---<br/>Well-rested<br/>Asking exploratory questions<br/>Slow interaction pace<br/>Morning/start of week]
        
        LoadDecision -->|High Load<br/>Stressed, Urgent| HighLoad[HIGH LOAD STATE<br/>---<br/>Quick clicks<br/>Scanning behavior<br/>Decision urgency<br/>End of day/week]
        
        LoadDecision -->|Critical Load<br/>Emergency| CriticalLoad[CRITICAL LOAD STATE<br/>---<br/>Extreme time pressure<br/>Crisis indicators<br/>Operational emergency<br/>Very short interactions]
    end
    
    subgraph "Adaptive Delivery Engine"
        LowLoad --> DetailedMode[DETAILED MODE<br/>---<br/>Comprehensive Analysis<br/>Multiple Scenarios<br/>Strategic Thinking<br/>Rich Visualizations<br/>Exploratory Questions]
        
        HighLoad --> SimplifiedMode[SIMPLIFIED MODE<br/>---<br/>Key Actions Only<br/>Clear Next Steps<br/>Defer Non-Urgent<br/>Minimal Context<br/>Quick Decisions]
        
        CriticalLoad --> EmergencyMode[EMERGENCY MODE<br/>---<br/>Single Best Action<br/>Hide All Non-Essential<br/>Immediate Response<br/>Supporting Info On-Demand<br/>Crisis Management]
    end
    
    subgraph "Content Formatting"
        DetailedMode --> DetailedFormat[Expanded Views<br/>Full Reasoning<br/>Tradeoff Analysis<br/>Long-Form Content<br/>Deep Dive Options]
        
        SimplifiedMode --> SimplifiedFormat[Bullet Points<br/>Action Checklist<br/>Summary Cards<br/>Progressive Disclosure<br/>Snooze Options]
        
        EmergencyMode --> EmergencyFormat[Single Alert<br/>One Recommendation<br/>Immediate Action<br/>Minimal Text<br/>Follow-up Later]
    end
    
    DetailedFormat --> Response[Cognitively-Appropriate<br/>Response Delivery]
    SimplifiedFormat --> Response
    EmergencyFormat --> Response
    
    Response --> User[User Interface]
    
    Response -.->|Interaction Outcome| Learning[(CLAGA Learning System)]
    Learning -.->|Pattern Recognition| StateAnalysis
    Learning -.->|User Preferences| DetailedMode
    Learning -.->|User Preferences| SimplifiedMode
    Learning -.->|User Preferences| EmergencyMode
    
    style LowLoad fill:#2ECC71,stroke:#27AE60,stroke-width:2px,color:#fff
    style HighLoad fill:#F39C12,stroke:#E67E22,stroke-width:2px,color:#fff
    style CriticalLoad fill:#E74C3C,stroke:#C0392B,stroke-width:2px,color:#fff
    style StateAnalysis fill:#3498DB,stroke:#2980B9,stroke-width:3px,color:#fff
    style Response fill:#9B59B6,stroke:#8E44AD,stroke-width:3px,color:#fff
    style Learning fill:#34495E,stroke:#2C3E50,stroke-width:2px,color:#fff

CLAGA Detection Mechanisms

Signal Collection Layer

1. Interaction Speed Signals

What CLAGAs Monitor:

Time between clicks
Scroll speed
Query entry pace
Page dwell time

Interpretation:

Slow, deliberate → Low cognitive load (exploring, learning)
Fast, jumping → High cognitive load (stressed, scanning)
Very rapid, erratic → Critical load (crisis mode)

Example:

Low Load Pattern:
- User opens dashboard at 9am
- Spends 3 minutes reading each section
- Types queries slowly, pauses to think
→ CLAGA: "User is in analytical mode, show detailed insights"

High Load Pattern:
- User opens dashboard at 6pm Friday
- Quick clicks through multiple pages
- Scans headlines, doesn't read details
→ CLAGA: "User is time-pressured, show action items only"

2. Query Complexity Signals

What CLAGAs Monitor:

Question type (exploratory vs. directive)
Query length and specificity
Follow-up question patterns

Interpretation:

"How does X work?" → Low load, wants to learn
"What should I do about X?" → High load, needs decision
"Fix X NOW" → Critical load, emergency

Example:

Low Load Query:
"Can you explain how the CAGA network evaluates alignment?"
→ CLAGA: Detailed explanation mode

High Load Query:
"Which AI opportunity should we implement first?"
→ CLAGA: Direct recommendation with brief rationale

Critical Load Query:
"Customer pipeline is broken, what NOW?"
→ CLAGA: Single action, defer analysis

3. Navigation Pattern Signals

What CLAGAs Monitor:

Page sequence
Back-button usage
Search vs. browse behavior
Abandoned actions

Interpretation:

Linear progression → Low load, following logical path
Jumping between sections → High load, seeking specific info
Repeated back-clicks → Very high load, can't find what's needed

4. Contextual Signals

What CLAGAs Monitor:

Time of day
Day of week
Calendar events (if integrated)
Recent system alerts

Interpretation:

Monday 9am → Generally lower load
Friday 5pm → Generally higher load
After system alert → Elevated load
During scheduled "focus time" → Lower load

5. Historical Pattern Signals

What CLAGAs Learn:

User's typical interaction patterns
Stress indicators specific to this user
Preferred information density
Response to different formats

Personalization:

User A Profile:
- Prefers detailed analysis even under pressure
- Rarely uses simplified mode
→ CLAGA: Higher threshold for simplification

User B Profile:
- Gets overwhelmed by too much information
- Frequently requests summaries
→ CLAGA: Lower threshold for simplification

Cognitive Load State Definitions

LOW LOAD STATE

Characteristics:

User is calm, exploratory
Has time for strategic thinking
Asking "how" and "why" questions
Slow, deliberate interactions

User Experience in This State:

Expanded detail views available
Rich visualizations shown
Multiple scenarios presented
Tradeoff analyses displayed
"Dig deeper" options offered
Educational content included

UI Adaptations:

Dashboard: Full intelligence overview
Recommendations: Complete reasoning shown
Visualizations: Detailed graphs, heat maps
Options: All alternatives presented
Documentation: Comprehensive explanations

Example Scenario:

Situation: User opens CosentriQ Monday morning, reviews weekly intelligence report

CLAGA Detection:
- Time: 9:00 AM Monday (low-stress time)
- Pace: Slow scrolling, 2-3 min per section
- Queries: "Explain the rationale behind this ranking"
→ Load State: LOW

Delivery Mode: DETAILED
- Show full CAGA analysis for each opportunity
- Display tradeoff comparisons
- Include scenario modeling options
- Offer deep-dive documentation
- Present educational content about methodology

HIGH LOAD STATE

Characteristics:

User is time-pressured
Needs decisions, not exploration
Quick scanning behavior
End of day/week timing

User Experience in This State:

Simplified action-focused views
Clear "next steps" highlighted
Non-urgent items hidden or deferrable
Summary cards instead of full analysis
"Snooze until tomorrow" options

UI Adaptations:

Dashboard: Action items only
Recommendations: Top 3 with brief rationale
Visualizations: Simplified charts, key metrics
Options: Pre-filtered to best choices
Documentation: Executive summaries

Example Scenario:

Situation: User opens CosentriQ Friday afternoon, needs to make decision before weekend

CLAGA Detection:
- Time: 4:30 PM Friday (high-stress time)
- Pace: Rapid clicking, scanning headlines
- Queries: "What's the top priority right now?"
→ Load State: HIGH

Delivery Mode: SIMPLIFIED
- Show top 3 recommendations only
- Brief 2-sentence rationale for each
- Clear "Approve" or "Defer" buttons
- Hide detailed analysis (available on-demand)
- Offer "Review Monday" option for non-urgent items

CRITICAL LOAD STATE

Characteristics:

User in crisis or emergency
Extreme time pressure
Operational incident in progress
Very short, urgent queries

User Experience in This State:

Single best recommendation
Minimal explanation
Immediate action focus
All supporting info hidden (available on-demand)
Follow-up deferred automatically

UI Adaptations:

Dashboard: Emergency alert mode
Recommendations: ONE clear action
Visualizations: None (unless critical)
Options: Single best path
Documentation: None shown (link provided)

Example Scenario:

Situation: Critical system failure, customer-facing workflow broken

CLAGA Detection:
- Context: System alert triggered 5 minutes ago
- Pace: User lands on page immediately, no browsing
- Query: "Production onboarding is DOWN, what do I do?"
→ Load State: CRITICAL

Delivery Mode: EMERGENCY
- Single recommendation: "Pause new customer onboarding immediately"
- One supporting action: "Notify existing customers of delay"
- Defer all analysis: "Full incident analysis available after resolution"
- Suppress all non-essential notifications
- Provide simple "Mark Resolved" button

Adaptive Content Examples

Same Intelligence, Different Delivery

Scenario: AI implementation recommendation for ticket categorization

LOW LOAD DELIVERY:

┌─────────────────────────────────────────────────────────┐
│ AI-Powered Ticket Categorization - Full Analysis       │
├─────────────────────────────────────────────────────────┤
│                                                         │
│ Recommendation: IMPLEMENT (Priority #3)                │
│ Composite Score: 78/100                                │
│                                                         │
│ Strategic Alignment (CAGA-A): 85/100                   │
│ → Directly supports customer experience goals          │
│ → Aligns with data-driven culture                      │
│ → Strengthens competitive positioning                  │
│                                                         │
│ Human Capacity (CAGA-H): 60/100                        │
│ → Team currently at 85% utilization                    │
│ → Requires 2-week training ramp-up                     │
│ → Medium change management needed                      │
│                                                         │
│ Technical Feasibility (CAGA-T): 70/100                 │
│ → CRM integration available                            │
│ → Moderate complexity (3 integration points)           │
│ → 6-8 week technical timeline                          │
│                                                         │
│ [View Full Analysis] [Compare Alternatives]            │
│ [See Implementation Roadmap] [Risk Assessment]         │
└─────────────────────────────────────────────────────────┘

HIGH LOAD DELIVERY:

┌─────────────────────────────────────────────────────────┐
│ ⭐ Recommended: AI Ticket Categorization                │
├─────────────────────────────────────────────────────────┤
│                                                         │
│ Why: Strong strategic fit, positive ROI                │
│                                                         │
│ When: Q2 implementation (after team training)          │
│                                                         │
│ Action needed:                                          │
│ ☐ Approve for Q2 budget                                │
│ ☐ Schedule team training (2 weeks)                     │
│ ☐ Review compliance requirements                       │
│                                                         │
│ [Approve] [Defer to Monday] [Details ↓]                │
└─────────────────────────────────────────────────────────┘

CRITICAL LOAD DELIVERY:

┌─────────────────────────────────────────────────────────┐
│ 🚨 ACTION REQUIRED                                      │
├─────────────────────────────────────────────────────────┤
│                                                         │
│ RECOMMENDED: Approve AI ticket categorization for Q2   │
│                                                         │
│ [Approve Now] [Review Later]                           │
│                                                         │
│ Full analysis available after resolution ↓             │
└─────────────────────────────────────────────────────────┘

CLAGA Learning System

Continuous Improvement

What CLAGAs Learn:

User-specific patterns
- When is this user typically stressed?
- What triggers cognitive load for them?
- What information density do they prefer?
Outcome tracking
- Did simplified mode help or frustrate?
- Was emergency mode appropriate?
- Did user override CLAGA suggestions?
Preference mapping
- Some users prefer detail even under pressure
- Some users prefer brevity always
- Context-specific preferences

Feedback Mechanisms

Explicit Feedback:

"Show more detail" button → User wants richer content
"Too much information" button → User wants simplification
Snooze/Defer actions → Confirms high cognitive load

Implicit Feedback:

User expands collapsed sections → Wants more detail
User quickly dismisses modals → Wants less interruption
User repeatedly clicks "Summary" → Prefers simplified

Personalization Over Time

User Journey Example:

Week 1:
- CLAGA uses default thresholds
- Learns user prefers more detail than average
- Adjusts LOW/HIGH load boundaries

Month 1:
- CLAGA identifies Friday afternoons = consistently high load
- Automatically simplifies Friday 3pm+ content
- User approves 90% of simplified suggestions

Month 3:
- CLAGA recognizes user stress patterns
- Pre-emptively simplifies during known high-stress periods
- User reports feeling "less overwhelmed"
- CLAGA model refined to user preferences

Technical Implementation

Detection Algorithm

Cognitive Load Score = weighted_sum(
    interaction_speed_score * 0.3,
    query_complexity_score * 0.2,
    navigation_pattern_score * 0.2,
    contextual_score * 0.15,
    historical_pattern_score * 0.15
)

if load_score < 0.3: state = LOW_LOAD
elif load_score < 0.7: state = HIGH_LOAD
else: state = CRITICAL_LOAD

Threshold Customization

Each user has personalized thresholds:

User A (prefers detail):
- LOW → HIGH threshold: 0.5 (higher than default)
- HIGH → CRITICAL threshold: 0.8

User B (easily overwhelmed):
- LOW → HIGH threshold: 0.2 (lower than default)
- HIGH → CRITICAL threshold: 0.6

Response Time Optimization

CLAGA detection runs in <50ms
Does not block content delivery
Can adjust mid-session if load state changes

Design Principles

Non-Intrusive: Detection happens passively, no explicit user input required
Reversible: Users can always request more/less detail
Personalized: Adapts to individual preferences and patterns
Context-Aware: Considers time, situation, historical patterns
Continuously Learning: Improves accuracy over time
Respectful: Assumes users are capable, just time-constrained

Research Basis

This approach is grounded in:

Cognitive Load Theory (Sweller, 1988)
Information Overload Research (Eppler & Mengis, 2004)
Adaptive Interfaces (Jameson, 2008)
Real-time Stress Detection (Hernandez et al., 2011)

File Information

Created: December 2025
Version: 2.0
Part of: OAI³ Framework Architecture Documentation
Related Diagrams:
- MIA Orchestration Flow
- CAGA Network Architecture
- Complete System Integration

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

CLAGA Adaptation Flow

Overview

Diagram

CLAGA Detection Mechanisms

Signal Collection Layer

1. Interaction Speed Signals

2. Query Complexity Signals

3. Navigation Pattern Signals

4. Contextual Signals

5. Historical Pattern Signals

Cognitive Load State Definitions

LOW LOAD STATE

HIGH LOAD STATE

CRITICAL LOAD STATE

Adaptive Content Examples

Same Intelligence, Different Delivery

CLAGA Learning System

Continuous Improvement

Feedback Mechanisms

Personalization Over Time

Technical Implementation

Detection Algorithm

Threshold Customization

Response Time Optimization

Design Principles

Research Basis

File Information

FilesExpand file tree

03_CLAGA_Adaptation_Flow.md

Latest commit

History

03_CLAGA_Adaptation_Flow.md

File metadata and controls

CLAGA Adaptation Flow

Overview

Diagram

CLAGA Detection Mechanisms

Signal Collection Layer

1. Interaction Speed Signals

2. Query Complexity Signals

3. Navigation Pattern Signals

4. Contextual Signals

5. Historical Pattern Signals

Cognitive Load State Definitions

LOW LOAD STATE

HIGH LOAD STATE

CRITICAL LOAD STATE

Adaptive Content Examples

Same Intelligence, Different Delivery

CLAGA Learning System

Continuous Improvement

Feedback Mechanisms

Personalization Over Time

Technical Implementation

Detection Algorithm

Threshold Customization

Response Time Optimization

Design Principles

Research Basis

File Information