Montis Coaching Intelligence Design

Montis uses a deterministic coaching pipeline that separates measurement, interpretation, operational decision-making, strategic governance, and report rendering.

This document describes the current production design and how the Montis Intelligence Stack fits together.

Detailed scientific and implementation behaviour for each intelligence layer is documented separately under Science & Methodology.

Why Montis Shows More Than Raw Metrics

Montis is not designed as a replacement dashboard for Intervals.icu.

Intervals.icu provides the underlying training data, metrics and calendar context.

Montis adds a governed interpretation layer on top of that data.

Its purpose is to answer:

• What does this data mean?
• How is the athlete responding?
• Is adaptation occurring?
• Is the current load being absorbed?
• What does the future plan imply?
• What should change next?

The value of Montis is therefore not simply the number of metrics displayed.

It is the relationship between them.

The Intelligence Stack progressively turns validated athlete data into physiological context, performance interpretation, adaptation state and a governed coaching decision.

In simple terms:

Intervals.icu provides the data.

Montis makes the data meaningful.

Data → Meaning → Adaptation → Decision

Coaching Intelligence Pipeline

VALIDATED TIER-1 / TIER-2 OUTPUTS
                │
                ▼
1. TRAINING LOAD
Applied stress and load pattern
                │
                ▼
2. PHYSIOLOGY RESPONSE
Autonomic, recovery and wellness response
                │
                ▼
3. PERFORMANCE INTELLIGENCE — PI
WDRM / ISDM / NDLI
                │
                ▼
Training State
load_accepting / recovery_priority
                │
                ▼
4. ENERGY SYSTEM PROGRESSION — ESPE
Power-curve adaptation state
                │
                ▼
5. ADE v2.21 BASE DECISION — CAN
Operational capacity and risk
                │
                ▼
PHASE + EVENT GOVERNANCE — SHOULD
Recovery, build, taper and event-form control
                │
                ▼
SEMANTIC OUTPUT
Actions / Guidance / Alignment / Event Context
                │
                ▼
URF v5.1 RENDERING
Governed report structure
                │
                ▼
LLM / APPLICATION PRESENTATION
Read-only interpretation

The Coaching Intelligence Pipeline operates only on validated outputs from the technical pipeline.

It does not allow the LLM to recompute canonical metrics, invent physiological states, or replace governed coaching decisions.

The Five Intelligence Layers

1 — Training Load

Training Load establishes the objective stress applied to the athlete.

It answers:

How much stress was applied?

This layer provides the load, volume, intensity and capacity context required by the downstream intelligence layers.

Detailed methodology:

Science → Training Load

---

2 — Physiology Response

Physiology Response evaluates how the athlete is responding to the applied stress.

It answers:

How is the body responding?

This layer resolves available autonomic, recovery, sleep, resting-HR, subjective and load-pressure evidence into physiological context.

The physiology state is an interpretation layer. It does not independently override Performance Intelligence, ADE, phase governance or final training guidance.

Detailed methodology:

Science → Physiology Response

---

3 — Performance Intelligence

Performance Intelligence evaluates how fitness is being expressed under current training stress.

It answers:

Is the athlete absorbing and expressing the applied load effectively?

Its current model contracts include:

• WDRM — anaerobic repeatability and W′ depletion behaviour.
• ISDM — durability and cardiovascular drift.
• NDLI — high-intensity density and execution characteristics.

Performance Intelligence also contributes to the consolidated Training State used by ADE.

For ADE, the operational state is simplified to:

load_accepting
recovery_priority

These states answer:

Can the athlete currently tolerate further training stress?

Detailed methodology:

Science → Performance Intelligence (PI)

---

4 — Energy System Progression

The Energy System Progression Engine (ESPE) evaluates longer-term change in performance capability.

It answers:

Is capability improving, remaining stable, or declining?

ESPE compares equivalent rolling power-curve windows and resolves changes across the principal performance systems.

Its outputs include:

• energy-system status;
• adaptation state;
• curve dynamics;
• power-model context;
• adaptation bias;
• system guidance.

The dashboard Adaptation Pulse is a presentation of ESPE output, not a separate adaptation model.

Detailed methodology:

Science → Energy System Progression (ESPE) & Phase Development

---

5 — Adaptive Decision Engine

ADE v2.21 is the operational decision layer.

It answers:

What can the athlete tolerate now?

ADE consumes governed upstream evidence including:

• operational state;
• forecast fatigue and load trend;
• HRV guardrail;
• ESPE adaptation state;
• target-event and taper context;
• supplementary nutrition context where confidence is sufficient.

ADE produces a pre-phase-governance base score and directive.

Detailed methodology:

Science → Adaptive Decision Engine (ADE)

Can vs Should

Montis deliberately preserves two different decisions.

ADE Base Decision — CAN

ADE = CAN

This represents immediate operational capacity.

Phase and Event Governance — SHOULD

Phase and event governance = SHOULD

Strategic governance considers:

• current and projected phase;
• recent fatigue/recovery requirements;
• planned load direction;
• taper requirements;
• target-event context;
• projected event form;
• controlled sharpening where appropriate.

Example:

Can: continue productive loading
Should: recover and consolidate adaptation

This is not a contradiction.

The first statement describes operational capacity.

The second describes what is strategically appropriate for the athlete's current phase or event objective.

Final Resolution

The final governed action can be classified as:

Semantic Output

The coaching pipeline exposes governed semantic sections such as:

training_volume
wellness
performance_intelligence
energy_system_progression
actions
training_guidance
decision_context
phase_alignment
event_targets
future_forecast
future_actions

The semantic layer preserves computed outputs and adds structured context for reporting and application use.

It does not recompute Performance Intelligence, ESPE or ADE.

Unified Reporting Framework — URF v5.1

URF v5.1 governs how validated Montis outputs are organised and rendered into reports.

It does not calculate the athlete's state or coaching decision.

The distinction is:

Intelligence Stack = analysis and decision
URF = governed reporting and presentation
Dashboard Horizon = time perspective

LLM Responsibility

The LLM is a read-only interpretation layer.

It may:

• explain the governed result;
• compare Can versus Should;
• present the training state clearly;
• summarise the reasons;
• suggest a calendar change through an explicit tool action.

It may not:

• recalculate canonical metrics;
• replace engine classifications;
• invent missing values;
• override the ADE result;
• create an alternative coaching directive;
• modify the training calendar without explicit execution.

Historical Report Safety

Current weekly guidance is tactical only when the report is current.

For stale historical weekly reports, Montis can:

• mark ADE output as historical_only;
• suppress live event-readiness guidance;
• remove current taper instructions;
• reframe the result as historical block analysis.

Historical output describes the state resolved at that time and must not be presented as today's coaching instruction.

How to Use the Help Manual

Use this page to understand how the complete Montis architecture fits together.

For detailed methodology, use the five Science sections:

• Science → Training Load
• Science → Physiology Response
• Science → Performance Intelligence (PI)
• Science → Energy System Progression (ESPE) & Phase Development
• Science → Adaptive Decision Engine (ADE)

For application-specific behaviour, use:

• Dashboard Horizons — how the stack is presented across time horizons.
• Wellness & Physiology — how wellness measurements are displayed and interpreted.
• Race Readiness — event, form, taper and readiness governance.
• Workouts & AI Builder — workout creation, planning and execution workflow.

Bottom Line

• Training Load establishes the applied stress.
• Physiology Response evaluates how the body is responding.
• Performance Intelligence evaluates whether that stress is being absorbed and expressed effectively.
• ESPE evaluates how capability is changing over time.
• ADE determines what the athlete can tolerate.
• Phase and event governance determine what the athlete should do.
• URF governs how the validated result is reported.
• The LLM explains the governed result without becoming the computational or prescriptive authority.