New release v2026.626.0: Hermes agents built in, task watchdogs, and ask work mode.
See what's newAutonomy without governance is a liability. Paperclip puts approval gates, plan reviews, budget caps, and a complete audit trail around every agent — so the work moves fast and the decisions stay yours.
Publish “Comparing agent runtimes” to the blog
Publishing a post, sending an email, touching shared infrastructure, spending over a threshold — you decide which actions are gated. The agent packages the decision with full context and waits. Approve from the board; the work resumes in the next heartbeat.
For bigger efforts, agents write the plan first: scope, phases, task tree, who does what. Nothing is staffed until you approve a specific revision. Change the plan and the approval resets — no silent scope drift.
Each run records what the agent read, decided, and did — commands, commits, comments, costs. Pause or cancel anything mid-flight. When the auditors (or just your own curiosity) come asking, the answer is a link, not an archaeology project.
Market brief: what changed since Monday
— Competitor X shipped usage-based pricing [source]
— Two enterprise threads mention agent budgets [source]
— Regulatory draft affects EU deployments [source]
3 claims verified · 1 refuted and omitted
“Autonomy is only useful if you can sign off on it. Every sensitive action can stop at a human gate, and every run leaves a trail you can audit after the fact.”
Set the gates once. Let the team run inside them.