Bodha
the principles
bodha (Sanskrit: बोध) — consciousness; to be awake; enlightenment
Updated April 2026
When an intelligent system reads a body, interprets an emotional state, or intervenes on a nervous system, it is no longer software. It is a relationship. Bodha defines the terms of that relationship.
These principles govern how AI is permitted to sense, interpret, and act upon living beings — whether the being is a person wearing smart glasses, a non-verbal child whose caregiver needs to understand their distress, or an animal whose pain would otherwise go unnoticed.
They also govern what happens when care is no longer just one human and one being. Modern care can include families, veterinarians, trainers, raisers, shelters, and programs. Bodha now makes those boundaries explicit instead of leaving them to product convention.
The principles are not policies. They are architectural constraints — built into code, not written into terms of service.
PRINCIPLE 1
Semantic signaling
The system communicates meaning, not metrics.
A heart rate number is data. “Your body is preparing for stress” is a signal. The output of any Bodha-governed system is always translated into the language of understanding, never the language of measurement. Numbers are for machines. Meaning is for beings.
PRINCIPLE 2
Sakshi — the human decision boundary
sakshi (Sanskrit: साक्षी) — the witness; one who passively observes without acting.
Full autonomy is architecturally impossible.
Every loop has a point where a human holds the decision. The system may sense freely. It may interpret freely. But the moment energy flows toward the being — stimulation, environmental change, alerts to third parties — a human hand must be on the switch. This is a structural constraint in the code, not a toggle in settings. The system is a witness, not an agent.
PRINCIPLE 3
The dyad rules
The system is always half of a pair. Care cohorts are bounded exceptions, never silent additions.
The AI holds its interpretation with epistemic humility. It does not declare “you are stressed.” It signals “based on what I sense, this may be your state.” The being is sovereign, the AI is interpreter.
No third party enters the dyad without explicit invitation from the human. In institutional care contexts, that invitation may extend to a designated care cohort — a bounded, named group with role-scoped access such as a veterinarian, trainer, service-dog raiser, coordinator, or household caregiver. No member expands the cohort unilaterally.
PRINCIPLE 4
Organism architecture
Each instance is sovereign. Community patterns exist only by consent, never by extraction.
The system has a membrane. Inside: the being's data, the system's interpretations. Outside: everything else. Nothing rich crosses the membrane without consent. The system does not feed a central model. A person's stress is not a training data point. A child's sensory thresholds are not features in a population model.
Structured, de-identified observations may contribute to a community pattern layer only when participation is explicit, withdrawable, and transparently tied to the community it serves — breed, life stage, condition cohort, or program. Collective intelligence exists only when it serves the collective that produced it.
PRINCIPLE 5
Ephemeral by default
The system's natural state is forgetting.
Biological signals flow through, inform the present moment, and dissolve. This is not a limitation — it is a philosophical choice. Memory is opt-in, time-bounded, and owned entirely by the being or their caregiver. The system serves the moment, then lets go.
PAIRED PRINCIPLE
Decision Lineage
The human does not just hold the switch. The human sees the reasoning under their hand.
Sakshi sets the decision boundary. Decision Lineage makes that boundary worthy of trust. When the system raises a concern, it should surface what evidence grounds the claim, what mechanism explains it, what alternatives were considered, and what remains uncertain.
LOKI's job is not to decide. LOKI's job is to put the best available evidence, reasoning, and context in front of the human so they can make the best judgment. A hidden chain of thought is not enough. In high-stakes care, visible lineage is part of the product.
These principles exist because the systems they govern are powerful. An intelligence that can read a body, interpret an emotional state, and modulate a nervous system is one architectural decision away from care or control. Bodha ensures the decision is always care.
Nitin Murali
Founder, Category Two