Atlas engine · v0.1.0

Guidance that shows its work.

Atlas is the future decision layer for ShaOS. It organizes evidence, identifies what is missing, and supports adaptation without pretending that an algorithm replaces athlete judgment.

Open private athlete portal

Public Atlas demo · synthetic data

See the evidence model without exposing the athlete.

Demonstration only
Sleep7h 42mSynthetic · Fresh
HRV58 msSynthetic · Fresh
Resting HR54 bpmSynthetic · Fresh
Readiness82%Synthetic derived metric
Strength volume18,420 lbsSynthetic loaded session
Evidence stateAvailableEvery value retains source and freshness

Privacy boundary

Public ShaOS and the Private Athlete Portal have different jobs.

Public ShaOS

Mission architecture, movement education, articles, benchmarks, and synthetic Atlas examples.

Private Athlete Portal

Apple Health, Hevy history, assessments, readiness, recovery, and athlete-specific decisions behind owner authentication.

01

Observe

Collect check-ins, training history, benchmark evidence, and field notes through stable records.

02

Explain

Show which evidence contributed, which evidence is stale, and which evidence is still missing.

03

Recommend

Support hold, adjust, recover, or progress decisions while keeping the athlete in control.

Readiness policy

Three evidence streams. No invented certainty.

Atlas does not convert missing evidence into zero. Readiness remains not measured until every required signal is available.

50%

Benchmark evidence

Weighted completion of project-and-phase benchmarks

30%

Recovery trend

Recent check-ins across sleep, soreness, stress, energy, and motivation

20%

Training consistency

Share of planned sessions completed or intentionally modified

Intelligence contract · v0.1.0

Recommendations are evidence-backed and reversible.

Atlas recommendations use normalized evidence, expose their rationale, and never replace athlete or clinician judgment.

  1. CollectWait when required normalized evidence is missing.
  2. ProceedSupport the planned mission when evidence stays inside policy thresholds.
  3. AdjustPreserve mission intent while managing elevated workload.
  4. RecoverPrioritize restoration when recovery indicators cross caution thresholds.

Guardrails

Blank is not zero

Missing data never becomes a poor score.

Private stays private

Operational health data does not appear on public pages.

Humans approve progression

A high score informs the decision; it does not make it automatically.

Policies are versioned

Changes to weights and thresholds remain reviewable and auditable.