Agents and agentic coding

Agentic coding is software that plans and writes software under instruction. Here is what it does well, where a senior engineer still decides the outcome, and how it changes what you should expect to pay for.

The agentic coding tools your developer is using, explained without the hype: what each one is for, who tends to pick which, and the questions that tell you whether it is being used well.

Bespoke software used to mean six figures and a season. Here is what agentic coding made cheap, what it left expensive, and how to buy custom software now without paying 2022 prices.

A fixed workflow follows rules. An agent uses judgment. Here is when each is right, how the risk differs, and why "an agent inside a workflow" is the pattern that works for most businesses.

An operations agent reads the inbox, checks your systems, and proposes what to do. Here is how to build one that earns trust one rung at a time, what tools to give it, and what to log.

MCP is the standard way for AI to use your business tools. Here is what it does, why it saves you from a pile of one-off integrations, the security questions it raises, and what to ask any vendor who says they support it.

An agent that can act needs the same discipline as a new employee with system access, applied more strictly. Here is the least-privilege setup: what to give it, what to withhold, and what to log.