Research
Research notes on autonomous intelligence under uncertainty.
NAO is developed through bounded research questions: how an agent perceives, reasons, preserves uncertainty, remembers prior state, evaluates outcomes and acts under governance.
Public notes reflect demonstrated work, active development and long-term research direction. They are not claims of human-like consciousness or guaranteed capability.
What is NAO?
Foundational positioning
A governed cognitive core, not a trading bot identity.
NAO Core is a generic cognitive architecture: it perceives, compares, reasons under uncertainty, remembers, revises and acts within explicit boundaries. EHUD++ is its coherence discipline — it evaluates factual coherence, systemic coherence, robustness and adaptability against uncertainty, contradictions and risk. Gold/XAUUSD is DomainAdapter_Gold — the first specialized environment plugged into that core. Bank, travel, shopping and personal assistance remain future domain directions, not products and not demonstrated transferability.
Governed autonomy
Demonstrated principle
Why belief is not authorization.
NAO separates observation, interpretation, proposal, authorization and action. A plausible interpretation can remain WAIT when evidence is incomplete, contradictory or too uncertain. More intelligence should not imply less governance.
BELIEF≠PROPOSAL≠AUTHORIZATION≠ACTION
Core and domain specialization
Architectural discipline
A domain should connect through an adapter, not redefine the core.
NAO is designed so that domain-specific data, rules and interfaces remain outside the general cognitive core whenever possible. A new domain should connect through an adapter rather than redefine the core.
If the domain disappeared, would the concept still be necessary to NAO?
YES → Core candidateNO → Domain Adapter
Architectural discipline — not a proof of successful multi-domain transfer.
Why Gold first?
First proving ground
A real proving ground for uncertainty.
Gold is not presented as proof of profitability or as NAO's final identity. It is the first environment used to test whether the architecture can remain coherent when information is incomplete, conditions change and mistakes have measurable consequences.
Research environment — not a performance claim.
In development
What comes next
Current research continues around memory, self-evaluation, outcome comparison, governed adaptation and the transfer of the core architecture to other complex environments.
NAO research increasingly evaluates capabilities by contribution to the agentic loop: perception, representation, decision, continuity, outcome evaluation and governance. A technically interesting capability does not automatically belong in the Core.