amux for Indie Hackers

Run a 5-10 agent engineering team from your phone. Ship features overnight, automate the boring parts, and build products that compound while you sleep. Updated July 2026.

amux is an open-source control plane for running an AI engineering team from a single dashboard or your phone. It launches, monitors, and self-heals dozens of Claude Code, Codex, or Gemini CLI sessions simultaneously. Built-in kanban board, notes, CRM, Gmail API, browser automation, and scheduler — everything an indie hacker needs to run a product without a team.

The Indie Hacker Advantage

The traditional indie hacker bottleneck is output velocity. You have a full backlog, a product that needs constant improvement, customers asking for features, support tickets piling up — and only one of you. Hiring feels premature when MRR is $500 and you're still finding product-market fit.

AI coding agents change the math. With 5-10 agents running overnight, you can ship a week's worth of feature work before you wake up. The agents don't sleep, don't ask for raises, and don't need onboarding. The cost to run them (typically $5-30/night on API pricing) is far less than a single engineering hour.

What amux adds is the infrastructure to make this reliable: a self-healing watchdog that restarts crashed agents, atomic task claiming so agents don't do duplicate work, a shared kanban board for tracking progress, and an iOS app for checking in from anywhere without opening a laptop.

A Typical Indie Hacker Week with amux

Sunday night setup (15 minutes)

Queue your week's feature backlog in the amux board. Write task cards — one per feature, bug, or improvement. Add context in the description. Assign priority. Set 5-8 agents running with YOLO mode enabled.

# Queue tasks for the week
amux board add "Add CSV export to the analytics dashboard" --priority high
amux board add "Fix mobile layout bug on the pricing page" --priority high
amux board add "Write unit tests for the billing module" --priority medium
amux board add "Update onboarding email sequence with new features" --priority medium

# Start 5 agents (stagger starts to avoid rate limit spikes)
for i in 1 2 3 4 5; do
  amux start "agent-$i" &
  sleep 8
done

Monday morning review (20 minutes)

Open the iOS app. Check which tasks moved to "done." Read the output of each agent in the peek panel. Review PRs that were opened overnight. Merge what's clean, leave feedback on what needs iteration. Move remaining tasks back to "doing" for day 2.

A realistic outcome for a well-prepared overnight run of 5 agents across 8 hours: 3-5 tasks completed, 1-2 needing review, 1-2 still in progress. You wake up to material progress on a full sprint's worth of work.

During the day (as needed)

Keep 2-3 agents running on lower-priority tasks during working hours. Check the board from your phone during lunch. Send task updates to specific sessions with a tap. The iOS app shows live output from each agent — you can follow along or ignore it entirely.

Beyond Coding: Running Your Business

Most indie hackers are not just building software — they're handling support, marketing, partnerships, and finances. amux's built-in tools cover more than coding:

Task amux tool How it works
Customer support Gmail API Agent reads inbox, drafts replies with your real signature, sends on approval (or auto-sends for common queries)
Market research Browser automation Agent visits competitor pages, pulls pricing, screenshots, summarizes changes to your board
User tracking CRM Log calls, notes, and follow-ups per contact; agents can read and update CRM records
Daily ops Scheduler Schedule recurring tasks: daily metrics pull, weekly summary post, monthly invoicing check
Content / docs Notes + coding agents Agent writes changelog, docs, and blog posts from recent commits and your product notes

Overnight Agent Workflow for Indie Hackers

The highest-leverage use of amux for indie hackers is the overnight run. Here's a proven setup:

  1. Write atomic tasks. Each task should be completable in 2-4 hours with clear acceptance criteria. "Add CSV export" is better than "improve the analytics page." The more specific the task, the higher the overnight success rate.
  2. Use separate git branches per agent. Register each session with a different working directory or branch to avoid conflicts. Each agent opens a PR from its branch — you review and merge in the morning.
  3. Enable YOLO mode. With --yolo, agents never block on tool-approval prompts. Essential for unattended runs.
  4. Set a CLAUDE.md with your project context. Agents work better with a project README, coding conventions, and "gotchas" document in the project root. One good CLAUDE.md means less back-and-forth clarification.
  5. Check in on the iOS app. The amux iOS app shows live agent output and board status from anywhere. You can send a quick correction message if an agent goes in the wrong direction — without opening a laptop.

The Cost Math for Indie Hackers

Common question: is AI agent coding cost-effective at indie scale?

Setup Cost Output
5 Claude Sonnet agents, 8h (API pricing) ~$15-40/night 3-6 PRs completed
5 Claude agents, Claude Max plan $200/month flat Unlimited runs included
5 Gemini Flash agents (docs/tests) ~$1-5/night High-volume mechanical tasks
1 human freelancer, 8h $400-800 1-3 tasks (with coordination overhead)

At $30/night for 5 agents, you're spending $900/month on AI engineering capacity that would cost $15,000-30,000/month in freelancer time. The quality difference is real — agents need more review and correction than senior engineers — but for an indie hacker willing to review output and iterate, the leverage is extraordinary.

Common Indie Hacker amux Setups

The Weekend Shipper

Queue 15-20 tasks on Friday night. Run 5-8 agents all weekend. Come back Monday to 8-12 completed PRs. Review and merge Monday morning, then spend the week on product decisions, marketing, and customer calls — not coding.

The Daily Autopilot

2-3 agents running 24/7 on the feature backlog. The scheduler posts a daily board summary every morning. Support emails are handled by an agent with Gmail API access. The founder checks in twice a day via iOS app, reviews PRs in the evening.

The Sprint Machine

For launching a new product or major feature: spin up 10+ agents for an intensive 48-hour sprint. Modularize the work, assign different modules to different agents, and parallelize everything that can be parallelized. The 48-hour sprint guide covers this in detail.

Getting Started

If you've never run AI coding agents before, the fastest path is:

  1. Clone amux and run ./install.sh (5 minutes)
  2. Register your project: amux register myapp --dir ~/Dev/myapp --yolo
  3. Start one agent and watch it work on a single task
  4. Scale to 5 agents once you understand the output quality and task description requirements

Or skip the setup entirely: the Concierge does the configuration for you and sets up your fleet from scratch in a managed onboarding session.

Start running your AI engineering team

amux is free, open-source, and runs on your existing hardware. No SaaS subscription, no per-seat fees — just AI agents working while you sleep.

git clone https://github.com/mixpeek/amux && cd amux && ./install.sh
amux register myapp --dir ~/Dev/myapp --yolo
amux start myapp
amux serve  # → https://localhost:8822
View on GitHub Concierge setup →

Related resources: amux for Solopreneurs · amux for Startup Founders · Overnight Agent Guide · Overnight SaaS MVP · Self-Healing Configuration