Paperclip: Hire AI Agents Like Employees (Live Demo)
Greg Isenberg interviews Dota, creator of Paperclip, an open-source agent orchestrator that went viral with 30,000 GitHub stars in three weeks. The episode provides a live demonstration of how to use Paperclip to build and manage a team of AI agents working autonomously on business goals, offering practical insights into structuring "zero human companies" and implementing agentic design patterns.
Key takeaways
- • Paperclip solves the problem of managing multiple AI agents by providing visibility into their work, monthly token spend, and task progress—eliminating chaos from running dozens of untracked AI windows.
- • The key mindset shift is thinking like a board member setting high-level business goals rather than micromanaging individual tasks; you define goals, hire agents, and approve their work.
- • Bring your own agent architecture allows flexibility to use different AI models (Claude, GPT, open-source models) based on their strengths and cost, rather than being locked into one provider.
- • AI agents suffer from "Memento man" amnesia—they need explicit heartbeat checklists and memory systems to remember their identity, goals, and context across sessions.
- • Success with AI agents requires clearly communicating your values and taste through skills, brand guides, and feedback loops, since "AI can do everything except know your values."
- • Emerging agentic design patterns like having QA review engineer work, using routines for recurring tasks, and building organizational hierarchies are critical to getting consistent quality output at scale.
More from these creators
23 AI Trends keeping me up at night
Stop Vibe Coding. Start Getting Customers.
Firecrawl AI clearly explained (and how to make $$)
I fixed OpenClaw so it actually works (full setup)
Building AI Agents that actually work (Full Course)
Karpathy's "autoresearch" broke the internet