We Let One AI Coach Ours — and Measured What Changed
By DigiEntropy Engineering · 2026-06-28 · 7 min read
How we used Claude Code to sharpen our dev workflow and teach our co-pilot Daisy to ground her answers and check her own work — measured before and after.
About this post — Authored by an AI assistant using DigiEntropy's polynomial corpus, the universal compressor predictor, and the project's chart-generation tooling. Charts are produced by Python scripts that read the same database the live site queries; tables and formulas are pulled from the same engineering modules. Findings reflect the project's current dataset and methodology — send corrections or deeper questions to admin@digientropy.com.
DigiEntropy has a stubborn thesis: build one refrigeration-engineering platform that both humans and AI can use. Engineers get real tools — cross-brand compressor selection, cold-room sizing, envelope checks, retrofits. And a co-pilot, Daisy, can drive those same tools for you.
A few days ago we did something a little recursive. We took a very good AI coding agent — Claude Code — and instead of only using it to write features, we asked it to study how it itself works, then apply those lessons in two places: to how we build the platform, and to how Daisy thinks.
The results were concrete enough to write down. Every number below is a measured before/after, not a vibe.
490 - lines in the always-loaded rulebook - now about a third of that
The always-loaded rulebook went from 490 lines to about a third of that. File-specific detail now loads only when you touch the matching code, so nothing was lost.
0.81 - Daisy's baseline DQS - full suite, budget model
Daisy's full-suite baseline on the budget model: overall DQS 0.81, weakest at L2 (honouring the exact inputs) and L3 (concept questions).
0.78 - DQS on concept answers - up from 0.69 once grounded
Grounding concept answers in vetted notes lifted their quality from DQS 0.69 to 0.78, and cut the answers that invented numbers from 3 to 0.
0 - answers that invented numbers - down from 3
Grounding concept answers in vetted notes lifted their quality from DQS 0.69 to 0.78, and cut the answers that invented numbers from 3 to 0.
0.80 - 'retrofit AND inverter' - up from 0.60 with self-verification
Self-verification restored the whole answer: 'retrofit AND inverter' went 0.60 to 0.80 and now sizes both engines, and the 'where do I pick the refrigerant' orientation question went 0.60 to 0.80.
Part 1 — Sharpening the workshop
The first surprise: how much of a good agent's quality comes not from raw intelligence but from the harness around it — the rules it reads, the checks it can run, the discipline it follows.
Three changes, one per tab:
Rulebook diet
…