Systems Engineering · Agent Architecture · AI-assisted Development

Aygent Jay

My long-term project for a shared architecture of specialized systems and AI agents.

The starting point is Ay Trade: a running trading system that I want to extend step by step with agents for market and trade evaluation. In the long term, I want to use the same foundation for research, coding, debugging and system management.

Long-term Vision

Three specialized directions

No single agent is meant to do everything. The systems should remain specialized while sharing information and capabilities where useful.

01

Trading & Capital

Ay Trade

The current trading core is intended to develop into an increasingly autonomous system that evaluates market opportunities from multiple perspectives and can eventually make a sustainable economic contribution.

02

Research & Information Quality

Ay Search

A research and knowledge system designed to compare sources, expose provenance, evidence and contradictions, and help people assess unreliable or manipulative information more effectively.

03

AI & System Work

Ay Core

A shared AI and agent foundation for coding, debugging, research, system maintenance and communication between specialized systems.

Current State

What is already running

The longer-term vision is built on working projects and live infrastructure.

Core System Active

Ay Trade

Market data, strategies, lifecycle management, validation, automation and controlled execution within a continuously running system.

View Ay Trade →
Reporting Experiment Active

Ay Report

An automatically generated daily market report inspired by institutional reporting – intentionally without LLM-generated report text.

Open Ay Report →
Visualization Experiment Beta

Ay Heatmap

An interactive liquidation heatmap based on Bitget market data – built as my own alternative to login- and paywall-restricted solutions.

Open Ay Heatmap →

Workflow

AI-assisted Systems & Prototyping

I use AI extensively for implementation, analysis and iteration. What matters to me is the loop between the intended goal, system behavior, real-world testing and the next improvement.

01 Problem & Goal
→
02 AI-assisted Implementation
→
03 Real-System Test
→
04 Review & Iteration
System Design Prompting Agent Workflows Automation Validation APIs Linux Technical Prototyping

Roadmap

From the foundation to Step 33

This overview only shows the larger phases. The technical roadmap itself is considerably more granular.

Steps 1–8 Done

Foundation

  • Trading Core
  • State & Lifecycle
  • Validation & Guards
Steps 9–17 Current

Reliability & Efficiency

  • Audits & Robustness
  • Storage & Runtime
  • Operational Reliability
Steps 18–23 Next

Consolidation & Interfaces

  • Consolidate modules
  • Agent-ready interfaces
  • Context & Integration
Steps 24–32 Planned

Agent & AI Foundations

  • Trade Evaluation Agents
  • Research & AI Layer
  • System Workflows
Step 33 Milestone

Re-Evaluation

  • Ay Trade Masterplan
  • Ay Search Masterplan
  • Ay Core Masterplan

Step 33 is deliberately not an endpoint. From there, the systems are intended to continue with separate masterplans based on the shared foundation developed up to that point.