› middle-management

the operating system for companies that run on AI

You're building with Claude Code. You've got agents, workflows, and projects running in parallel. middle-management keeps them governed, coordinated, and visible — so you can scale without losing control.

what it does

governance and coordination that Claude Code doesn't have

structured task contracts

Every piece of work has an objective, acceptance criteria, and constraints. Agents don't guess what "done" looks like — it's defined before they start.

dispatch and orchestration

middle-management queues work, assigns it to specialist agents, and tracks outcomes. You see what's running, what's finished, and what's blocked — across every project.

review and quality assurance

Built-in challenger workflows catch problems before they compound. Work is reviewed against acceptance criteria, not vibes.

persistent operational state

A real database — not markdown files. Task history, session outcomes, and coordination state survive across sessions. Nothing gets lost between conversations.

how it works

one conversation to get started

1

tell middle-management what you're working on

Projects, teams, priorities — whatever your current state is. middle-management creates structured task contracts for each piece of work.

2

middle-management orchestrates the work

It queues tasks, dispatches agents, tracks progress, and runs quality checks. You get a standup every morning with what happened, what's next, and what needs your attention.

3

you make the calls — middle-management handles the rest

Strategic decisions stay with you. Sequencing, tracking, dispatching, and quality assurance run through the system. Scale up without scaling your overhead.

the problem

AI-first is easy to start. It's hard to scale.

You started with one Claude Code session. Then two. Now you've got agents running across multiple projects and no single place to see what's happening, what succeeded, and what went sideways.

Claude Code gives you powerful building blocks — sub-agents, scheduled tasks, MCP servers, plugins. What it doesn't give you is governance. No persistent task queue. No outcome tracking. No quality assurance. No structured work contracts. You're building your own operating system out of markdown files and hope.

middle-management is the product layer that sits on top of Claude Code and turns it into something you can actually run a company on. Structured contracts. Persistent state. Dispatch and review. Morning standups with what happened overnight.

Free. Open source. MIT licensed. Built for people who are already in this deep.

what's built

this is real. here's what exists today.

middle-management is in active development and early testing. Here's what's already working:

  • SQLite database — persistent operational state across all sessions
  • Task contracts — structured objectives, acceptance criteria, and constraints for every piece of work
  • Dispatch system — queues work, assigns specialist agents, tracks completion
  • Challenger/review workflow — built-in quality assurance before work is marked done
  • Morning standup — automated daily briefing on progress, blockers, and next actions
  • Browser dashboard — real-time view of queue, agents, tasks, and project health
  • Multi-project coordination — dependencies, sequencing, and cross-project visibility

Not yet ready for general release. Currently in early testing with a small group of AI-native startup builders.

get updates

stay in the loop

We'll let you know when middle-management is ready for onboarding. Sign up for updates.

contact

say hi

Running AI across your company and feeling the coordination pain? Let's talk.

hello@middle-management.ai