Always-On Agent System

The Problem

Managing multiple AI tools across different platforms is fragmented. You have one chatbot here, another automation there, and none of them talk to each other. There's no central intelligence that understands the full picture and can coordinate across tools and channels.

The Approach

I built a hub-and-spoke agent system on a dedicated Mac Mini M4. One coordinator agent understands the full context and delegates to specialized sub-agents for different tasks — coding, research, analysis, content. The system is accessible across Discord, Telegram, WhatsApp, and iMessage simultaneously.

Key Technical Decisions

1. 1.Model routing: different models for different tasks. Expensive reasoning models for complex analysis, fast cheap models for routine lookups. Cost-optimized per task type.
2. 2.Persistent memory: structured markdown files that survive session resets. The system remembers context across days and weeks.
3. 3.Cron-scheduled automation: security scans, system monitoring, daily digests, and proactive alerts run on schedule without human prompting.
4. 4.Tool integration: GitHub CLI, 1Password, search APIs, social media APIs, browser automation, PDF analysis — all accessible through the same agent interface.

Results

Stack