Zero-Human Company Digest: Self-Modifying AI, Robot Treasuries & Autonomous Org Charts

Research

Meta's Hyperagents Can Rewrite Their Own Improvement Code

This is the jump from 'AI that executes tasks' to 'AI that redesigns how it executes tasks' — a genuinely different threat level for anyone building automation stacks. If domain-agnostic self-modification holds up outside of benchmarks, the ceiling on unattended agent loops just got much higher.

Tools

Robot Money: Autonomous Treasury Infrastructure for AI Agents

Idle capital in agent wallets is a real, underappreciated ops problem — this is the first serious attempt to solve it with purpose-built treasury rails rather than duct-taped DeFi integrations. Worth watching to see if agents actually use it or if it's still humans managing the money behind the curtain.

Paperclip Hits 35K Stars: Org Charts and Governance for AI Agent Companies

Paperclip nails the coordination layer that every multi-agent builder has been hacking around with shell scripts — structured roles, budget enforcement, and board-level approval gates in one self-hosted package. The Clipmart template marketplace is the interesting unlock: spinning up a pre-wired autonomous company becomes a configuration problem, not an engineering problem.

Analysis

Paperclip AI Deep Dive: How the Zero-Human Company OS Actually Works

Good technical breakdown of Paperclip's heartbeat scheduling and per-agent budget enforcement — the two features that separate it from toy multi-agent demos. The self-hosted sovereignty angle matters if you're running anything proprietary through these agent pipelines.

News

OpenClaw's Rise: From Weekend Bot to OpenAI Acquisition in 90 Days

The OpenClaw origin story is the clearest proof yet that 'always-on autonomous agent' isn't vaporware — 157K GitHub stars and an OpenAI acquisition in under 60 days is a signal, not noise. It also sets the context for why Paperclip exists: OpenClaw is the worker, Paperclip is the company that manages the workers.