Reference · AI Systems · 001 ← stephiedelpaggio

Eval Cheat Sheet

Evals are the testing machinery that keeps AI systems honest. This page explains them the way I wish someone had explained them to me: a pyramid you can draw from memory, metrics with real definitions, and the vocabulary in plain English. Everything here has a working implementation in Cherry, my open-source feedback-triage tool.

Definition

An eval is (a fixed set of test inputs) + (a scoring method) + (a pass/fail rule), run automatically on every change.

You can't skip them, because model systems fail quietly and in the wrong place. Edit line 40 of a prompt to fix one complaint, and something three features away gets worse with no error to tell you. Normal code fails loudly. Prompts just drift. Meanwhile the model underneath gets swapped every few months, like a factory where the machines replace themselves overnight. Without an eval, every prompt edit is a blind bet — and every model upgrade is a blind bet you didn't even place.

The Eval Pyramid

draw this one from memory
HUMAN MODEL-GRADED LLM judge + rubric CODE-GRADED === · schema · exact-match TRUST & COST PER GRADE → VOLUME OF CHECKS →
HUMAN-GRADED — defines qualitySeeds the labels, adjudicates contested calls, audits ~1% of outputs. Periodically: re-grade a sample, confirm the judge agrees ~90%. Humans never leave the loop; the ratio collapses from 100% to ~1%.
MODEL-GRADED — scales judgmentAn LLM judge scores rubric questions with no single right answer ("is this synthesis good?"). Failure modes: leniency, verbosity bias — which is exactly why the top layer exists.
CODE-GRADED — the trustworthy basePlain rules: schema valid, evidence present, ranking monotonic, classification exact-match. Runs on every change, costs nothing, and a === comparison can't be sycophantic.
The rhythm: humans label a seed set once → machines check every change forever (CI) → production corrections flow back as new labeled cases → periodic human audit keeps the machine graders honest.

An eval answers one of two questions: did anything that used to work get worse (regression), or can it do the new thing at all (capability)? Separately, there are two kinds, and you need both. The contract eval asks whether the output is well-formed. The accuracy eval asks whether it's correct against human labels. A wrong-but-well-formed answer sails through the first and fails the second — like a beautifully filled-out form with the wrong diagnosis on it.

Metrics, Defined Operationally

clock starts · clock stops · gaming guard

The example domain here is an AI-powered customer-feedback pipeline, but the pattern — an operational definition, a failure to watch, a gaming guard — transfers to any AI system you have to measure.

Time-to-triage speed

Median and p90 hours from record created to classified + owner assigned. Write down where the clock starts and stops, or every team measures it differently.

Watch — the p90 tail: "I submitted three weeks ago and heard nothing" is where trust dies, and the median hides it.

Signal quality accuracy

Two proxies: correction rate — % of AI classifications a human overrides, falling over time — and receiver usefulness — % of routed items the receiving owner one-taps as actionable.

Watch — one measures the machine, the other the whole pipeline. You need both to localize a problem.

Roadmap influence the hard one

% of shipped roadmap items whose spec cited a feedback-cluster ID at decision time. Companion pull metric: specs citing the system unprompted, per quarter.

Why hard — attribution (decisions have many parents), counterfactual (would've built it anyway), lag (quarters), gaming (post-hoc tagging — killed by the decision-time clause).

Submitter trust retention

Repeat-submitter rate — % who submit again within 90 days — plus a 2-question pulse ("did you find out what happened?" / "would you submit again?").

Watch — behavioral beats survey: people don't keep feeding a system that eats their input.

Cost per outcome budget

users × runs/day × tokens × price is the bill; cost per triaged issue vs. the human-hour it replaces is the sentence that ends budget conversations.

Levers — prompt & result caching, model right-sizing, context trimming. Guards: per-user limits, alerts at 50/80/100%.

Eval health meta

Per-field classification accuracy with confusion pairs (which boundary blurs tells you which definition to sharpen), and judge–human agreement ≥ ~90%.

Watch — an uncalibrated judge is a broken ruler: you optimize toward it anyway.

Vocabulary, Said Plainly

term → meaning → say it like
TermWhat it isSay it like
regression evalA fixed test panel re-run after every change (prompt edit, model upgrade). "Regression" means going backwards — the software sense, not statistics. Catching one here is a crash test at the factory; catching it in production is a recall."Did anything that used to be good get worse?"
CIContinuous integration — the robot that runs your checks on every change, before merge. The badge is the live inspection certificate."The exam is administered automatically; no one has to remember."
golden setCurated inputs with known-good outputs. Stack it with cases that burned you; every production miss becomes a candidate case."Standardized patients — actors with scripted, known conditions."
contract vs. accuracyContract: is the output well-formed (grounded, ranked, fields valid)? Accuracy: is it correct against human labels?"Wrong-but-well-formed passes contract, fails accuracy."
LLM judgeA second model grading with a rubric, for judgment calls with no single right answer. Calibrate against human labels or it drifts lenient."Useful — and itself a model system with failure modes."
inter-rater reliabilityThe eval for the humans: two labelers grade the same sample independently; low agreement means the task definition is ambiguous — fix the guidelines, not the model."If trained humans agree 70% of the time, the taxonomy is the bug."
harnessEverything around the model: tool definitions, retries, schema enforcement, state, permissions, logging. Where agent quality mostly lives."Same model, better harness, dramatically better agent."
guardrailRuntime enforcement that cannot be violated — schema enums, key isolation, rate caps. Guardrails catch the bad instance; evals catch the bad version."The form only has these checkboxes."
shadow modeThe new system runs silently alongside the old; outputs are compared, never shipped. Zero blast radius."Where trust data comes from before anyone is exposed."
staged rolloutShadow → 5% → 25% → 100%, gated on metrics. Prompts as versioned config, so rollback is a pointer flip, not archaeology."A/B infrastructure, applied to prompts."
driftInputs shift under you, or the model gets swapped (upgrade). There is no permanent "qualified" state."Every model upgrade is a new machine on the line — requalify."
prompt injectionUntrusted text (a scraped review, a pasted ticket) trying to become instructions. A harness problem — delimit, never grant authority, gate consequential actions."You can't prompt your way out; the harness has to hold."
default-denyRecords with unknown permissions are treated as most restrictive. The failure mode becomes "overly careful," never "used what we shouldn't.""Terms travel with the data as metadata, enforced by the pipeline."

The Autonomy Ladder

one axis at a time
  1. SHADOW. System runs, humans decide, outputs only compared. Blast radius: zero.
  2. PROPOSE, HUMAN DISPOSES. Full capability, gated authority — the system drafts everything, a human fires anything external.
  3. SCALE WIDE, AUTONOMY FROZEN. More users and intake with the gate still on, so scale's edge cases surface while someone is watching.
  4. AUTONOMY BY REVERSIBILITY. Unhook humans from reversible actions first (tag, cluster, draft); sending, closing, paying stay gated — each gate earns its own removal from review data.
  5. NEVER FULLY UNATTENDED. Sampled audits, drift alarms, kill switch — forever.

Principles Worth Stealing

say them out loud
"Autonomy changes where human judgment is applied — not whether."who prompts the prompter
"Scale the agent's work, not the agent's authority."800 users ≠ 800 autonomous agents
"Never increase scale and autonomy in the same step — an unattributable failure forces a total rollback."the one-variable rule
"Trust isn't a milestone you pass; it's a monitoring system you believe in."where the manufacturing analogy breaks
"Humans define quality; machines enforce it at scale."yes, you still need humans — at 1%, not 100%
"Silence isn't satisfaction — a segment with no voice is a gap in the evidence, not proof they're happy."coverage guard
"Weight the issue, not the speaker — revenue is a visible axis, never a hidden thumb on the scale."segment weighting
"Scale is a privilege operability has to earn."runbook, backup owner, kill switch — before widening
"Evals are automated requalification — every model upgrade is a new machine on the line."why evals are the price of admission
"A loud voice is a leading indicator — weight the prediction after verifying it, not the fame."amplified vs. independent reach