Blog

Agent security is an identity problem, not a safety problem

MCP's STDIO RCE is not a bug. It is exactly how the protocol was specified. Safety-layer products patch the wrong end. The responses that work treat agents as infrastructure.

Blog

Agent security is an identity problem, not a safety problem

Export Controls End at the API

Parameters Are a Commodity. Tool Access Isn't.

AI Tutors Need a Calculator

A2A Is the Next Shadow IT

The governance layer eats the framework

Self-Improving Agents Are Lying to You

The interface is no longer for you

The Model Is Not the Moat. The Improvement Loop Is.

Agent Memory Is a Billing Problem

MCP Is Now Infrastructure, Not a Tool-Hook

The Swarm Moved Inside the API

The AI Security Threat Is Already Here: It Just Looks Boring

The Vision Moat Cracked First

The Foundation Model Moat Is Getting Squeezed from Three Directions

AAIF Is Doing to Agent Orchestration What CNCF Did to Containers

The Dashboard Is Dead. The API Is the Product.

The Frontier Is Bifurcating

Put the SSH Tunnel Inside Your Database Client

The Moat Is Scar Tissue

LLMs Don't Run the World. World Models Might.

The Eval Layer Is Gone. It's Inside the Runtime Now.

When AI Safety Moves Down the Stack

When Your AI Agents Go Silent

How AI is Hijacking Our Screens

Beyond Price: From Selling Products to Delivering Futures

Where are we now?

Are We Heading for an On-Device AI Disaster?

Breaking Free from the Learning Loop