Nous Ergon
νους εργον — intelligence at work
Origin
Nous (νους) is mind, intellect, the capacity for reason. Ergon (εργον) is work, deed, function — the same root as English ergonomics and energy. The name frames what the project is: agentic intelligence applied to a measurable, continuously verifiable problem. The work — the ergon — is what's on display.
Thesis
Build an experimentation harness for systematic equity strategies — multi-agent research, machine-learning prediction, risk-gated execution, weekly self-tuning — and instrument every decision it makes end-to-end. The orchestration pattern consists of six modules collaborating through S3 contracts, three Step Function pipelines on a fixed cadence, and an autonomous feedback loop that writes optimized parameters back into the system.
The harness is the durable artifact; alpha capture against the S&P 500 is the first experiment inside it. The system is currently in Phase 2: Reliability + Measurability buildout — making the instrument trustworthy enough that Phase 3 can turn alpha tuning on. Alpha is tracked, but not optimized, until measurement is trustworthy.
Built by
Brian McMahon. Single-developer project, in development since March 2026; Claude Code (Anthropic's LLM coding assistant) is the active collaborator on the implementation pass. Each module's repo is public.
Nous Ergon is a harness for agentic-orchestration research where different architectures, models, and patterns can be tested against measured baselines. Equities were chosen because financial data is abundant, decisions are unambiguous, and outcomes are continuously verifiable.