Insights
Notes on strategy, positioning, go-to-market, and emerging technology, from building AI-native operations in production.
Featured · AI OperationsAn Agent Governance Framework in Three Layers
Who is accountable when your agents run? A three-layer agent governance framework that gates the risky work without slowing the routine.
Read the article →
How to Configure Execution Policies in Paperclip
An execution policy sits between an agent calling done and Paperclip recording done. Here's how to configure review and approval gates at the task level.
Paperclip Alternatives and Competitors, by Category
The AI agent market is not one category. Orchestrators, runtimes, developer frameworks, and managed clouds each solve a different problem at a different layer. This guide maps the field before the comparison starts.
Why Paperclip vs Hermes Is the Wrong Question
People searching for 'Paperclip vs Hermes' are asking the wrong question. They sit at different levels of the stack. Here is the orchestrator-vs-runtime distinction and when Hermes runs as a per-agent adapter inside Paperclip.
Which Paperclip adapter for which agents?
Paperclip lets you run each agent on a different adapter: claude_local, hermes_local, codex_local, or opencode_local. Here is how to choose the right one.
How to Harden Your Agent Runtime to Fail Closed
Fail-open agent runtimes execute whatever they are handed. Here are the hardening layers that make an agent runtime fail closed by default.
When an Agent Says a Task Is Done but It Isn't
Your agent marks a task done, but the work isn't there. Here's why false-success happens, how to diagnose it in your logs, and how review gates fix it.
Why AI Agent Token Costs Keep Climbing, and How to Cut the Overhead
A Hermes analysis found 73% of every agent API call is fixed overhead. Here's why it's structural, why caching only gets you part way, and what actually reduces your agent token bill.
Why Your Agents Get Marked Failed When They Succeeded
Up to a third of agent runs get marked as failed while the work completes fine. Here's why run-level and task-state diverge, and the config that stops it.
Agent Sandbox Security
Agent sandbox security in two dials: a default-deny egress proxy for agents and a dangerous-command approval gate at the sandbox boundary. The config that stops a credential leak at the network edge.
Securing Your Agent Runtime
Harden your agent runtime before production: sandbox isolation, credential isolation, egress control, and dangerous-command approval gates, the four controls that make an agent safe to leave unattended.
How to Set Up a Multi-Agent Team
Setting up multi-agent orchestration for real work? Here is the role-based pattern (coder, design, test, supervisor) and where a human stays in the loop.
What to Do When Your AI Agent Gets Stuck
Your AI agent keeps stalling, and restarting never lasts. The three root causes of a stuck agent, and the recoverable setup that fixes it.
How to use Paperclip Blueprints and wire the company it gives you
Paperclip Blueprints builds an agent company from a brief in minutes. Here's the by-hand wiring after import (skills, secrets, models, budgets, heartbeats) that turns an imported company into one that actually runs.
How to Set Up an AI Agent Company
How to set up an AI agent company: an operator's guide to org design, agent mandates, budgets, governance, and recovery when an agent gets stuck.
Grolea Rewrite: From growth consulting to AI-native operating systems
Why Grolea is moving from traditional growth consulting to build AI-native operating systems, and why I am betting on an open-source AI stack.
Paperclip Blueprints: Generate Paperclip companies from a markdown brief
An open-source CLI that turns a markdown brief into a complete, schema-validated Paperclip company bundle, with the platform's best practices built in. Why I built it, and how to try it.