REDDIT · REDDIT// 13d agoPRODUCT UPDATE

Autonomous AI Trading Lab Adds Episodic Memory

The project now stores bugs, decisions, and principles in a JSON episodic-memory log instead of a RAG or vector-store layer, and RayoBot plus Darwin consult it before acting. It also added performance-based capital allocation and a Tivoli Constitution that kills products without traction in 30 days or sales in 60.

// ANALYSIS

This is less a trading bot than a self-governing agent stack, and that governance layer is becoming the real product.

–A JSON log is a cheap, explicit memory layer that preserves context without forcing a premature retrieval system.
–Species Capital Allocation makes the system adaptive: mean_reversion's 2.02 PF earns 1.5x capital instead of a flat split.
–The Tivoli Constitution is a useful antidote to optimism, because product survival now depends on measurable traction.
–The downside is obvious: hard kill rules can prune slow-burn ideas before they have time to compound.

// TAGS

agentautomationreasoningautonomous-ai-trading-lab

DISCOVERED

13d ago

2026-03-29

PUBLISHED

13d ago

2026-03-29

RELEVANCE

8/ 10

AUTHOR

piratastuertos