The Agent Office blog
Notes on running coding agents on real tickets.
Posts
-
What Matt Pocock gets right about coding agents
Matt Pocock stopped writing code by hand and runs agents all day. His four lessons on telling agents what you care about, the codebase as their environment, effort and trust, and how to use them.
-
What is agentic coding? (vs vibe coding)
Agentic coding means giving an AI agent a task it can plan, code and test. Compare it with vibe coding and see a practical workflow with human review.
-
Codex vs Claude Code (2026): which to use, and how to run both
Codex vs Claude Code in 2026, compared on models, pricing, where each runs and what each is good at. Plus how to put both on real tickets and compare the pull requests.
-
Agentic engineering: how to manage AI agents like a team
What agentic engineering is, how it differs from writing every line yourself, and the practices that make managing AI coding agents work like managing a team.
-
How to run multiple AI coding agents at once (without the chaos)
How to run Claude Code parallel agents with git worktrees, where the manual setup falls apart at three agents, and what a ticket board changes.
-
Why we built The Agent Office
The story behind The Agent Office, from two lazy engineers who wanted to get more done, to a coding agent on every ticket.
-
Best AI coding tools in 2026 (free and paid)
The best AI coding tools in 2026, grouped into terminal, IDE and cloud agents, with what each is best at, what it costs and what you get free. Plus how to use several at once.
-
Put a Claude Code agent on every Linear issue
How to connect Linear to The Agent Office so each assigned issue gets its own Claude Code agent, and the PR and summary land back on the issue when you approve.
open the office
For Apple Silicon Macs with a coding agent installed. Windows coming soon





