9 min read
AI sessions are contractors. Always-on agents are hires.
A chat or a coding agent session ends when the task is done, when the answer is given or the PR is merged. Grok Bot and OpenAI's dots stay. That changes what you have to give them.
Posts
Blog
7 min read
Eight months of an MCP server for Db2 for i
In January the server looked up table definitions so the AI would stop guessing. At 2.12 it checks its SQL against the IBM i, reads your business definitions, and lets colleagues sign in as themselves.
5 min read
AI-assisted ERP document generation
ERP documents look simple. Underneath, they're conditional logic puzzles that have resisted modernization for decades.
9 min read
Subagents replaced my /code-review command
Rules tell agents what to do. Subagents verify they actually did it.
5 min read
Agent context is the new technical debt
Git tracks what changed. With AI-generated code, the reasoning behind it matters more. Entire makes that context durable.
6 min read
Debugging with AI is a diagnosis problem
Stop asking your AI to fix bugs. Ask it to investigate.
6 min read
Let AI supervise itself
Wire AI into a development workflow that catches mistakes automatically.
5 min read
Plan before implementing
Speed without direction is expensive iteration. How planning with AI changes the dynamic.
2 min read
An MCP server for DB2 for i
I open-sourced an MCP server that lets Claude, Cursor, and other AI tools talk directly to DB2 for i.
3 min read
Use reference documentation tools with AI agents
Stop babysitting the model with pasted snippets. Pull the right docs into context automatically.
5 min read
Why AI works better on existing codebases
The assumption is that AI-assisted coding works best on fresh projects. After a year of experience, I think the opposite is true.
8 min read
From fragmented code to consistent output with AI rules
Most developers use AI without structured guidance. Here's how capability-specific rules change everything.
