Blog

The shift from generative AI to agentic AI, and why the "swivel chair" workflow is finally coming to an end.

How to build a "padded room" for your agent and stop worrying about indirect attacks. Learn the anatomy of prompt injection and shift from prevention by instruction to prevention by architecture.

Google just restricted paying AI subscribers for connecting through OpenClaw. Here's why provider lock-in is the biggest unspoken risk in the agent ecosystem — and what to do about it.

A candid look at the small scoped run that helped me trust agents: keeping our LLM catalog up to date without giving them the keys to production.

Quick recap of the February release: Discord Components v2, nested sub-agents, cron webhooks, and a long list of security hardening wins across the stack.

Here's exactly how we set up an open-source AI agent as an always-on ops triage layer — and eliminated overnight alert fatigue in six weeks.

OpenClaw hit 149k GitHub stars in under three months. The capability is impressive. The security posture? Still catching up. Here's where the cracks are showing.

Most people think they are using AI to automate their lives, but they are actually just using a faster typewriter. Here's why autonomous agents change everything.

Why we let an AI agent handle our expenses (and why you should too). Here's how teams actually use OpenClaw — from CI/CD monitoring to expense management.