Back to Portfolio
Agent Operations

Control Bridge

An agent control tower. One screen to dispatch, monitor, and govern every AI agent running across an operation. Built for the moment when "we have an AI initiative" turns into "we have twenty agents and nobody knows what they are doing".

FastAPI + SSE Docker Production

93-Second Walkthrough

93s 1280x720 Real product screenshots Originally posted on LinkedIn, 14 Apr 2026

Why I Built It

I have spent 11 years running mining service operations. The thing that breaks an operation is never the missing tool. It is the missing visibility. The same pattern shows up in AI rollouts: agents spawn faster than oversight, work happens in terminals nobody is watching, and the people accountable for outcomes have no live picture of what their fleet of agents is actually doing. Control Bridge is the operations console I would want if I were running an AI program at scale.

What It Does

Module 01
Agent Roster

Every agent is registered with a defined role, scope, and budget. Spawn, pause, retire from one screen. No more orphaned background processes burning tokens.

Module 02
Live Telemetry

Server-Sent Events stream every tool call, every decision, every state change as it happens. Operators see what an agent is doing without waiting for a log dump.

Module 03
Heartbeat Loop

A scheduled orchestrator runs nine phases on a cycle: scan, plan, dispatch, verify, learn. Every cycle leaves a save point. Every save point is reviewable.

Module 04
Human-in-the-Loop

Approval gates on the actions that matter. Destructive operations, external messages, and budget thresholds route through a human before they execute. Trust through traceability.

Technology Stack

FastAPI Server-Sent Events Docker Claude Code MCP Python Remotion TypeScript