Blog

How We Know an AI Agent Is Actually Good: Eval Harnesses and LLM-as-Judge

The difference between an agent that demos well and one you can put in front of customers is measurement. Here's how we score AI agent quality — eval harnesses, LLM-as-judge, and regression tests.

When the AI You Trusted Becomes the Door Left Open: The Vercel Breach

Vercel, the platform behind Next.js, was breached last month after an attacker exploited a single over-permissioned AI tool connected to a corporate Google account. The intrusion didn't start at Vercel — it started with a Roblox cheat download. Here's the full attack chain and what it means for any team running on modern cloud infrastructure.

The Next Wave of AI Coding: What Comes After Vibe Coding and Agents

Vibe coding and agentic tools are already table stakes. The real shift happening now — Spec-Driven Development, orchestrated multi-agent pipelines, and autonomous engineering orgs — will compress weeks of work into hours and redefine what it means to build software. Here's what's actually next and how to position your team for it.

We Taught Our Website to Think. It Won't Shut Up Now.

The Future of Commerce: When AI Builds Your Interface in Real-Time

Transform Your Space with Custom 3D Printed Solutions: From Mood Lighting to Pet Portraits

Using ComfyUI and Stable Diffusion for Exterior Design: Your New AI Architectural Assistant

SDXL Lightning generates AI images as fast as you can type.

Experimenting with the New & Powerful Anthropic Claude API

AutoGen: Changing the Game With Generative AI.

Using CrewAI agents to plan my next trip.

NextJS deployed to Vercel: 404 page not found