Skip to lesson
Exit
The Interview Loop1 / 2

1 min lesson

AI column / agent step

Use the example in "AI column / agent step" to explain the main idea in plain words.

Step 1 of 2

AI column / agent step: classify and personalize, but keep it grounded and idempotent
// Clay-style AI column or agent step: qualify + draft, with a guardrail
{
  "input": { "company": "{{company}}", "signal": "{{page_viewed}}", "firmographics": "{{enriched}}" },
  "task": "Classify ICP fit (A/B/C) from firmographics + signal. If A or B, draft one personalized opener grounded ONLY in the provided context. If C, return action=nurture and no draft.",
  "rules": [
    "Never invent facts not present in input (no hallucinated headcount/funding).",
    "Return strict JSON: { tier, action, reason, draft? } so downstream routing can branch.",
    "Same input must yield the same routing decision (idempotent); cache by company id."
  ]
}
Interview move

Demo with discipline. Walk the slice end to end, then say what you deliberately cut and why, the approach you rejected (for example, parallel enrichment because cost outran the match-rate gain) and how it generalizes to a new program by swapping the trigger. Defending those decisions is half the grade.

The AI-for-its-own-sake trap

Bolting an LLM onto a step that a formula or a lookup handles better reads as weak judgment, not AI-native impact. Reach for AI to classify fuzzy inputs, personalize at scale or let an agent take an action, and use deterministic logic everywhere it's cheaper and more reliable.