Autonomous Intelligence Engines

Self-monitoring systems that proactively detect risks, generate insights, and plan corrective actions — no user intervention required.

Overview

The autonomous intelligence layer contains 15 engines that model different aspects of cognitive and behavioral health. Unlike standard CRUD services, these engines maintain rolling statistical windows, classify lifecycle phases, predict future states, and generate proactive interventions.

🧠 AttentionDebtService

lib/core/services/attention_debt_service.dart

Models cognitive overhead as financial debt — unresolved items accrue cognitive interest over time using compound interest formulas. When total debt exceeds capacity, the system declares cognitive bankruptcy and triggers emergency triage.

Core Concepts

• Debt Item — any deferred decision, postponed task, or unresolved mental load
• Cognitive Interest — baseCost × (1 + dailyRate)^daysOpen — grows exponentially
• Postponement Penalty — each postponement adds 20% multiplier to effective cost
• Debt Score — 0–100 scale mapped from total debt vs. 500-unit capacity via sigmoid
• Bankruptcy Threshold — score ≥ 80 triggers emergency triage mode

Debt Categories & Interest Rates

Urgency Classification

Based on the ratio of effectiveCost / baseCost:

Sprint System

The service autonomously recommends sprint types based on current debt score:

Key Methods

Usage Example

🚀 MomentumEngineService

lib/core/services/momentum_engine_service.dart

Tracks task/habit/goal completion velocity (completions per day) over rolling windows, classifies momentum into lifecycle phases, detects blockers, and generates adaptive micro-nudges to sustain productive flow.

Momentum Phases

Key Concepts

• Velocity — rolling completion rate over configurable windows (default: 7/14/30 days)
• Acceleration — change in velocity between consecutive windows (positive = speeding up)
• Blocker Detection — pattern analysis to identify what's causing momentum loss
• Micro-Nudges — context-aware, adaptive suggestions calibrated to current phase

Key Methods

🐸 DriftDetectorService

lib/core/services/drift_detector_service.dart

The "boiling frog" detector — monitors gradual lifestyle regression across multiple metrics using linear regression over rolling windows. Detects slow negative changes that humans don't notice, predicts tipping points, and correlates cross-metric slides.

Drift Severity Levels

How It Works

1. Data Collection — aggregates daily metric values from other services (sleep quality, mood, productivity, etc.)
2. Rolling Regression — fits linear regression over configurable windows to detect slope direction
3. Tipping Point Prediction — extrapolates when a metric will cross a danger threshold at current rate
4. Cross-Metric Correlation — detects when multiple metrics drift together (e.g., sleep down → mood down → productivity down)
5. Recovery Nudges — generates targeted suggestions based on which metrics are sliding and their correlations

Key Methods

🔬 BehavioralFingerprintService

lib/core/services/behavioral_fingerprint_service.dart

Creates a multi-dimensional behavioral signature from daily patterns across 8 measurable axes and detects when the user deviates significantly from their baseline self using Mahalanobis-inspired distance metrics.

Behavioral Dimensions

Identity Phases

Key Methods

🔥 BurnoutDetectorService

lib/core/services/burnout_detector_service.dart

Multi-signal burnout risk analysis engine that aggregates stress indicators from across the app, computes resilience scores, detects warning patterns, and generates proactive recovery recommendations before burnout hits.

Risk Levels

Signal Sources

• Sleep Quality — declining sleep duration or quality trends
• Mood Patterns — persistent low mood or emotional flatness
• Productivity Changes — declining output despite same or increased effort
• Social Withdrawal — reduced contact tracker activity
• Habit Disruption — broken streaks across multiple trackers
• Energy Levels — sustained low energy readings
• Screen Time — excessive or erratic screen time patterns

Key Methods

⚡ WillpowerBudgetService

lib/core/services/willpower_budget_service.dart

Models daily willpower as a finite, depletable resource based on ego depletion theory. Starts at 100 each day, drained by cognitive demands (decisions, resistance, focus blocks), and partially restored by strategic recovery actions. Predicts decision fatigue windows and rates how well the user manages cognitive resources.

Willpower Zones

Demand & Recovery Model

• Demands — decisions, resisting temptation, sustained focus, context switches, emotional regulation
• Recovery — rest breaks, meals, exercise, meditation, nature exposure, sleep
• Fatigue Forecast — predicts when budget will cross each zone boundary at current drain rate
• Strategic Score — rates how well the user allocates high-demand tasks to high-budget periods

Key Methods

🌅 RitualEngineService

lib/core/services/ritual_engine_service.dart

Tracks daily routine execution patterns, learns optimal timing windows via historical analysis, detects 6 types of disruptions with severity scoring, and generates context-aware micro-adjustments to improve ritual adherence.

Rhythm States

Key Concepts

• Timing Score (0–100) — how close each execution was to the learned optimal window
• Chain Health (0–100) — overall daily ritual sequence scoring
• Disruption Types — 6 classified disruption patterns (schedule shift, motivation loss, external blocker, fatigue, overload, travel)
• Micro-Adjustments — small, actionable timing/sequence changes based on recent data

Key Methods

🔮 RegretMinimizationService

lib/core/services/regret_minimization_service.dart

Inspired by Bezos's regret minimization framework — records decisions with context and stakes, tracks outcomes over time, detects regret patterns (action vs. inaction), identifies recurring cognitive biases, and distills personalized decision-making wisdom. The ultimate question: "Will 80-year-old me regret NOT doing this?"

Decision Domains

Decisions are categorized across 8 life domains: Career, Relationships, Health, Finance, Education, Lifestyle, Creative, and Other. Each domain builds its own regret profile over time.

Analysis Pipeline

1. Decision Recording — log decisions with context, stakes, emotional state, and alternatives considered
2. Outcome Tracking — revisit decisions after configurable time periods to record actual outcomes
3. Regret Classification — categorize outcomes as regret of action (did it, wish I hadn't) or regret of inaction (didn't do it, wish I had)
4. Bias Detection — identify recurring cognitive biases from decision patterns (status quo bias, loss aversion, sunk cost, etc.)
5. Wisdom Distillation — generate personalized decision-making principles from your history

Key Methods

🛡️ StreakGuardianService

lib/core/services/streak_guardian_service.dart

Cross-tracker streak risk aggregator that monitors streaks from habits, workouts, meditation, water intake, sleep, and event activity. Detects at-risk streaks before they break and generates rescue strategies.

Risk Levels

Key Methods

🧮 DecisionFatigueService

lib/core/services/decision_fatigue_service.dart

Tracks decision-making patterns throughout the day, detects cognitive degradation, estimates remaining decision capacity, identifies peak quality windows, and proactively recommends batching or deferring decisions when fatigue is predicted.

Core Concepts

• Decision Event — each choice recorded with weight, category, and outcome
• Quality Score — estimated decision quality that drops as capacity depletes
• Capacity — daily decision budget (100 units) that depletes with each choice weighted by cognitive cost
• Peak Window — historical time slots when decision quality was highest
• Batch Suggestion — recommendation to group similar-category decisions together

Decision Weights & Cognitive Cost

Fatigue Levels

Decision Categories

Decisions are grouped by category for pattern analysis: Scheduling, Financial, Social, Health, Work, Creative, Personal. The engine detects which categories cluster together and suggests batching windows.

Key Methods

🌊 StressCascadeEngineService

lib/core/services/stress_cascade_engine_service.dart

Models how stress in one life domain cascades into others, tracks resilience buffers, detects approaching tipping points, and forecasts recovery trajectories. Uses a 6-domain stress propagation model.

Stress Domains

7 Sub-Engines

1. Stress Source Tracker — logs stressors with severity and domain classification
2. Cascade Simulator — models inter-domain stress propagation using weighted adjacency
3. Resilience Scorer — measures coping capacity via recovery rate analysis
4. Tipping Point Detector — identifies approaching breakdown thresholds per domain
5. Buffer Analyzer — tracks stress buffers (sleep, exercise, social support) and depletion rates
6. Recovery Forecaster — predicts time-to-recovery per domain based on historical patterns
7. Insight Generator — ranked actionable recommendations for stress management

Key Methods

🎯 FocusEntropyEngineService

lib/core/services/focus_entropy_engine_service.dart

Uses Shannon entropy (H = −Σ p log₂ p) to quantify how scattered attention is across life domains. Tracks context-switching costs, identifies deep-work blocks, scores flow-state quality, and forecasts focus trajectory via linear regression.

Focus Grades

7 Sub-Engines

1. Activity Logger — records focus sessions with domain, timestamp, duration
2. Entropy Calculator — computes Shannon entropy on domain time distribution
3. Context Switch Counter — counts domain transitions and estimates switching cost
4. Deep Work Detector — identifies uninterrupted ≥25-minute single-domain sessions
5. Flow State Scorer — composite 0–100 score from deep-work ratio and entropy
6. Focus Forecast — linear-regression trend on daily entropy values
7. Insight Generator — ranked actionable recommendations for focus improvement

Key Concepts

• Deep Work Block — any uninterrupted focus session ≥ 25 minutes on a single domain
• Context Switch Cost — estimated 15–23 minutes of reduced productivity per switch (based on UC Irvine research)
• Flow State — composite of deep-work ratio, low entropy, and sustained duration
• Focus Forecast — projects entropy trend forward to predict upcoming fragmentation risk

🦁 ChronotypeOptimizerService

lib/core/services/chronotype_optimizer_service.dart

Analyzes activity timing patterns to detect the user's natural chronotype, identifies peak performance windows for different task types, and provides personalized time-of-day recommendations for schedule optimization. Uses kernel density estimation on 24-hour energy curves.

Chronotypes

Task Types

7 Sub-Engines

1. Activity Clock Logger — records activities with timestamps and task types
2. Circadian Profile Builder — constructs 24-hour energy curve via kernel density estimation
3. Chronotype Classifier — classifies user as Lion / Bear / Wolf / Dolphin
4. Peak Window Detector — identifies optimal time windows per task type
5. Schedule Alignment Scorer — scores current schedule vs. optimal alignment (0–100)
6. Drift Detector — monitors circadian rhythm shifts over time
7. Insight Generator — ranked actionable scheduling recommendations

🤝 SocialCapitalEngineService

lib/core/services/social_capital_engine_service.dart

Autonomous relationship network health analyzer. Tracks social investment patterns, connection quality, reciprocity balance, and network diversity. Detects relationship neglect and provides proactive suggestions for maintaining social capital.

Key Concepts

• Social Capital Score — composite network health metric (0–100) combining diversity, reciprocity, and engagement
• Reciprocity Balance — ratio of investment given vs. received per relationship
• Connection Decay — relationships lose strength over time without interaction, modeled as exponential decay
• Network Diversity — how many distinct social circles the user maintains (work, family, friends, community)
• Investment Nudges — proactive suggestions when high-value relationships show decay signals

Relationship Tiers

🔧 FrictionJournalService

lib/core/services/friction_journal_service.dart

Autonomous micro-frustration tracker. Categorizes and patterns recurring friction points in daily life, calculates cumulative time and energy costs, and generates elimination strategies. Surfaces the "death by a thousand cuts" problem that humans typically ignore.

Key Concepts

• Friction Entry — a recorded frustration with category, severity, time cost, and context
• Pattern Detection — identifies recurring friction (same source, same time, same trigger)
• Cumulative Cost — aggregates minutes and energy spent on repeated micro-frustrations
• Elimination Strategy — ranked recommendations: automate, delegate, eliminate, accept
• Friction Score — per-category score combining frequency × severity × time cost

Friction Categories