Graphite pencil illustration of the Archmage of Analysis Paralysis, a wizard buried in comparison matrices

The Archmage never met an AI tool they couldn't spend another week evaluating. Before the team can adopt a coding assistant, the Archmage needs to benchmark it against four alternatives, run it on three representative codebases, and produce a comparison matrix that belongs in a peer-reviewed journal.

Symptom
Output is exclusively preparatory: evaluation documents, comparison spreadsheets, proof-of-concept repos. Calendar full of meetings about tools rather than meetings using tools. When pressed for a recommendation, they hedge.
Why it matters
The Archmage's cost is pure delay, but delay compounds. While they evaluate, the team doesn't build shared practices. When adoption finally happens, there's no institutional knowledge, no conventions, no guardrails.
What the chapter gives you
How to set decision deadlines with teeth, why most AI tool choices are low-stakes and reversible, and the bounded evaluation role that turns the Archmage into an asset.

Parent Class

From Volume 1 of The AI Developer's Field Guide

Read the full chapter in The AI Developer's Field Guide.

Recognize this one in your codebase?

The book has the full chapter, the symptoms, and the interventions.