In Service
of Life
An ecological intelligence platform that listens to the Web of Life, synthesises signals from multiple data layers, and supports humans in unlearning extractive patterns and co-learning regenerative ones.
What is on your mind?
Index
Platform overview. Principles. Navigation hub for all Stigmergy surfaces and documentation.
You are here →Diagnostic
Live system health tests. Verify Worker, KV, R2, Workers AI, CORS, biosignal, and clock endpoints independently.
Run tests →Architecture
Complete system design reference. All 16 sections — infrastructure, engines, endpoints, and ethical principles.
Read docs →AI That Listens to Life
Stigmergy takes its name from the biological principle where individual agents leave traces that guide collective behaviour — without central control. Ants deposit pheromones. The colony navigates without a navigator. This platform scales that principle to planetary ecological intelligence.
BM25 · RAG · Steward Network
A four-step pipeline: BM25 search finds relevant corpus documents, AI synthesises a grounded answer, citations are surfaced with direct links, and energy cost is disclosed. Every response is traceable. No hallucinations beyond the corpus.
Cloudflare Edge · 50 Endpoints
Built on Cloudflare Workers, KV, R2, and Workers AI. The single worker at api.ayu.earth handles all 50 endpoints. No servers. No containers. Edge-deployed globally.
Planet → Steward → AI
The planet is the ultimate authority. Stewards — humans with place knowledge — hold primary decision-making power. AI listens, synthesises, and surfaces patterns. It never prescribes. It never acts autonomously on behalf of stewards.
Life-in-the-Loop
Human-in-the-loop governs the intervention decision. Life-in-the-loop governs the outcome. A daily cron measures biosignal delta at each intervention's coordinates 30 days later — improving, worsening, or neutral — without asking the steward to self-report. The living system determines its own outcome.
Accounting for the Observer
Human-in-the-loop is itself perturbation. The Human Exceptionalism Guard flags when intervention frequency co-occurs with worsening signals. Actor Alignment Scores update automatically on bad outcomes. The trajectory: living system signals as authority — not human interpretation of them.