Continuity + context
Institutional memory, lived-experience records, decision memory, current-state context, corrections, preferences, and proposed learning.
OMOS • O-H-I Twin • Continuity • Human Authority
Oru’Valen™
Learns from lived reality.
Oru’Valen is the personalized O-H-I Twin and continuity/context layer. It is designed to preserve history, decisions, reasoning, outcomes, corrections, and current-state context so OMOS can reason with continuity while consequential execution remains governed through the Human Gate and ACC.
The canonical responsibility split
Oru supplies approved continuity and context. OMOS performs governed reasoning, Council synthesis, the Human Gate, Decision Records, persistence, and audit. ACC is the authorized execution control plane for approved external actions.
Governed reasoning
Distillation, alignment, Council review, synthesis, Human Gate, Decision Records, provenance, persistence, and audit history.
Authorized execution
Approved tools, workflows, agents, connectors, deployments, verification, external actions, and execution outcome reporting.
Human input / authorized evidence → Oru context snapshot → OMOS governed reasoning → Council / synthesis → Human Gate → Decision Record → ACC authorized execution → verification / audit → outcome → proposed Oru learning → human-approved memory/current-state update
Oru’Valen™ Lived Experience Layer
A useful twin must learn more than documents. The Lived Experience Layer models what happened, the context in which it happened, what decision was made, why it was made, what action followed, what outcome occurred, and what lesson should be proposed for future continuity.
Life Events
Material events, meetings, moves, wins, losses, commitments, business developments, and other user-approved real-world context.
Decision Memory
Problems, alternatives, constraints, decisions, stated reasoning, actions, and measured outcomes.
Behavior + Routine
Authorized evidence about recurring workflows, bottlenecks, actual execution patterns, and repeated operational choices.
Relationship Context
Factual, access-controlled records of relevant people, roles, interactions, commitments, and history—without speculative profiling.
Economic Reality
Permissioned evidence about what actually produces or consumes resources. Sensitive financial information remains restricted.
Preferences + Judgment
Repeated approvals, rejections, corrections, and choices that indicate durable operating preferences.
Lessons + Operating Rules
Proposed lessons derived from evidence and outcomes. The underlying events remain preserved and the lesson requires appropriate approval before becoming durable current state.
Fact is not inference
Oru must preserve epistemic boundaries. An observation, a user statement, an inference, and a prediction are different record classes and must never be silently collapsed into one another.
Verified event or record
Something established by an approved source, direct record, or verified external result.
Explicitly stated
What the human said, wanted, believed, intended, corrected, or instructed at that time.
Pattern interpretation
A reasoned interpretation based on evidence. It remains an inference until confirmed or superseded.
Expected future choice
A probabilistic forecast of what the human may prefer or decide—not a fact and never autonomous authority.
Correction rule: a direct human correction has higher authority over a prior model inference about that human. Historical records remain preserved; verified newer state may supersede earlier current-state assumptions.
Privacy, consent, and authority
Designed boundary
- Authorized sources only.
- No claim of background surveillance.
- Restricted memory requires access controls.
- Provenance, timestamp, source, confidence, privacy class, and approval state travel with records.
- Consequential external actions remain Human-Gate and ACC controlled.
Current maturity
- This page and machine profile define the integration contract.
- Caller-supplied Oru context and governed OMOS records may exist independently of this page.
- Automatic memory retrieval, automatic memory injection, complete outcome ingestion, and autonomous current-state editing are not represented here as Production capabilities.
- Deployment must be proven separately from repository implementation.