Writing Code Got Cheap. Owning It Didn't.
An agent built a genuinely hard part of our product: a tender-document ingestion pipeline with a job queue, retries, multi-format parsing and embeddings for search. I didn't write a line. Owning it, reviewing it hard enough to trust it, was the cost AI didn't make cheaper.
01
AI & Industry
Opus 4.6 Just Got 1M Tokens for Free. Here's What to Do With It.
Anthropic quietly flipped the switch on March 13. Claude Opus 4.6 now defaults to a 1M token context window for Claude Max subscribers. Here's what actually changes in your workflow.
02
Developer Tools
Stop Guessing: Test Your LLM Prompts with Promptfoo
You wouldn't ship code without tests. So why ship prompts that way? Promptfoo gives your prompts a proper test suite — catch regressions, compare models, and deploy with confidence.