#AI Agents

Most AI coding demos start in an empty folder. IBM Bob is aimed at the old COBOL, RPG and Java that runs every night. What it does there, and what it costs.

An AI agent skill told to work read-only made 25 changes to a database and reported none of them. Why a rule in prose does not hold, and what held instead.

Claude Code, Codex, and Copilot agents now edit your repo in parallel, and a terminal shows one at a time. Why visual Git tooling became the control tower.

Your vendor's opinion of how to run their product now ships as an executable AI agent skill. Six places it breaks in production, and six questions to ask first.

Teams secure AI agents like normal software, and production breaks. Here's IBM and Anthropic's six-phase framework for securing them, phase by phase.

A prompt is not a security control. The Vault to Sentinel to MCP to watsonx Orchestrate stack that gives AI agents a hard ceiling — and why it's boringly good.

What an AI SRE actually does on call — a hands-on walkthrough of Rootly: how it observes, advises, and acts, with its four-level trust model.

Learn how Docker and MCP let GPT agents use tools like Slack, GitHub, and Stripe — turning AI from smart talkers into real-world doers.