I run 6 AI-powered businesses from a single Vultr VPS. Each business needs different types of agents - some handle customer calls, others process documents, and a few manage data workflows. The challenge? Coordinating all these agents without chaos. My solution: a Supabase queue table that feeds tasks to Claude-powered dispatchers. Here's exactly how I built it. The Queue Table Structure First, I created a simple queue table in Supabase: CREATE TABLE task_queue ( id SERIAL PRIMARY KEY , business_id VARCHAR ( 50 ) NOT NULL , task_type VARCHAR ( 100 ) NOT NULL , payload JSONB NOT NULL , status VARCHAR ( 20 ) DEFAULT 'pending' , priority INTEGER DEFAULT 5 , created_at TIMESTAMP DEFAULT NOW (), assigned_at TIMESTAMP , completed_at TIMESTAMP ); CREATE INDEX idx_queue_status_priority ON task_queue ( status , priority DESC ); Enter fullscreen mode Exit fullscreen mode The business_id field maps to my 6 different companies. Task types include "customer_call", "document_review", "data_sync", and "lead_qualification".…