Skip to content
View coreyepstein's full-sized avatar

Highlights

  • Pro

Block or report coreyepstein

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
coreyepstein/README.md
$ whoami

Corey Epstein — founder, builder, person who taught AI agents to do his job so he could do more of it.

I run multiple companies from a single terminal. Not because I'm efficient — because I built a system that is.

I don't use AI to write code faster. I use it to run entire operations: shipping products, managing email, writing content, analyzing data, deploying to production. Agents that actually work, not demos that look good in a tweet.

"There's a better way. Find it."

The system

I built HQ — a personal operating system where AI agents are employees, not assistants.

HQ/
├── companies/          # 4 isolated business contexts (credentials never cross)
├── workers/            # 30+ specialized AI agents
│   ├── dev-team/       # 16 workers: architect → backend → frontend → QA → deploy
│   ├── content-team/   # 5 workers: brand → sales → product → legal → publish
│   └── ops/            # CFO, CMO, analyst, security scanner
├── projects/           # PRD-driven execution (write spec → agents build it)
├── knowledge/          # Searchable context per company, per domain
└── .claude/commands/   # 22 slash commands that orchestrate everything

Each company has its own credentials, knowledge base, deploy targets, and worker roster. I switch contexts by switching directories. The agents handle the rest.


How it works

The Ralph Loop is the core execution model. Named after Ralph Wiggum — because the best systems are simple enough that even Ralph could run them.

  ┌─────────────────────────────────────────────┐
  │                                             │
  │   Write PRD  →  Agents execute  →  Ship     │
  │       ↑                              │      │
  │       └──── Learn + improve ─────────┘      │
  │                                             │
  └─────────────────────────────────────────────┘

Write a spec. Go to sleep. Wake up to working software. The agents pick tasks, write code, run tests, fix failures, and loop until the PRD is done. Each cycle captures learnings that make the next cycle better.

This isn't theoretical. It's how I ship production code across all my companies.


What I've open-sourced

Project What it does
hq-starter-kit The full HQ framework — clone it, make it yours
ralph-methodology The autonomous coding loop documentation
advanced-gmail-mcp Gmail MCP server — multi-account email from the terminal
advanced-slack-mcp Slack MCP server — multi-workspace messaging
ralph-planner Collaborative PRD planning with AI assistance
hq-architecture Interactive visual guide to the HQ system
screenshotpath Screenshot → clipboard path (tiny macOS utility)

The stack

Languages    TypeScript · Node.js · SQL
Frameworks   Next.js · React · Tailwind · Playwright
AI           Claude Code · MCP Protocol · qmd (semantic search)
Infra        Vercel · Neon (Postgres) · DynamoDB · Resend
Mail/Comms   Gmail MCP · Slack MCP · Post-Bridge API

Background

CEO & Founder at Voyage (SMS + AI for e-commerce). Chief Design Officer at Abacus (enterprise AI for regulated industries). Building Indigo AI. MBA from UCLA Anderson. Ex-Deloitte. Based in Boulder, CO.

I've been building products for 15 years. The last two have been the most productive — because I stopped doing the work and started designing systems that do it for me.


X   LinkedIn   Website

Pinned Loading

  1. hq-starter-kit hq-starter-kit Public

    Personal OS framework for orchestrating AI workers. Built on Ralph methodology.

    TypeScript 12 2

  2. ralph-methodology ralph-methodology Public

    Ship working code while you sleep - The Ralph Wiggum Loop autonomous AI coding methodology

    6

  3. advanced-gmail-mcp advanced-gmail-mcp Public

    Gmail MCP server for Claude Code — multi-account email management

    TypeScript

  4. advanced-slack-mcp advanced-slack-mcp Public

    Slack MCP server for Claude Code — multi-workspace messaging, channels, and threads

    TypeScript

  5. indigoai-us/hq indigoai-us/hq Public

    Personal OS for AI Workers — install with npx create-hq

    TypeScript 3

  6. indigoai-us/hq-docs-app indigoai-us/hq-docs-app Public

    Indigo Docs — Your HQ, beautifully browsable. Tauri 2 + React 19 desktop app for navigating knowledge bases.

    TypeScript 1