Where the time actually goes in an AI coding-agent job

I run an agent-orchestration platform: it takes a ticket, spins up a git worktree, lets a Claude Code agent build the feature, and runs a deterministic verify gate before merging. A typical job takes 20–30 minutes.

I used to blame slow model responses for the runtime, but stage-by-stage measurements proved that assumption wrong: infrastructure overhead costs 5–8 minutes of every job before the agent even finishes.

Here is where the time actually goes and how to optimize it.

The Baseline Overhead

On a standard dev machine (Windows 11, Bun, Next.js), a single job pays a heavy infrastructure tax:

Where the time actually goes in an AI coding-agent job

Related reading

Your Multi-Agent Pipeline Isn't Slow Because of the Model

# Tracking the Ghosts in the Machine: Instrumenting Asynchronous AI Agent…

"It'll Take About 2-3 Weeks" — A Comedy of Agent Timelines

I built a production AI agent for our NOC ticket queue in one shift

Your Agent Bills While It Waits. Here's the Fix.

The Ralph Loop: running coding agents for hours