The Agent Tax: Who Pays When Nobody's in Charge?

Analysis

MIT Paper: Agent Societies Need Individuality, Not Just Swarms

This research pushes back on the 'just throw more agents at it' model, arguing that true autonomy requires distinct agent identities and goals. For zero-human companies, it means your 'workforce' might need an org chart with real roles, not just a homogenous blob of LLMs.

New Benchmark: Trust and Safety in Multi-Agent Systems

You can't audit what you can't measure. This paper introduces a framework for quantifying trust in agent teams, directly addressing the 'who's checking the checkers' problem that haunts autonomous orgs.

YC: The Economics of Running an Agent-First Company

Y Combinator is starting to talk about the real cost structures and operational metrics for startups built around autonomous agents. This shifts the conversation from 'can we?' to 'does it pencil out?'—a crucial filter for hype.

Tools

AutoGPT Pushes Production Hardening, Not Just New Features

The latest release focuses on reliability and error handling—boring stuff that actually matters when you're trusting an agent with your treasury. This is the unglamorous work that separates a demo from a deployable headless stack.

WalletConnect Integrates Directly with Agent Frameworks

This is the plumbing for autonomous treasuries: agents can now natively sign transactions without a human in the loop. It's a critical step for any zero-human company trying to manage its own capital, but it also means the liability for a bad transaction has no human to blame.