Skip to content

Instantly share code, notes, and snippets.

@23maverick23
Created March 23, 2026 20:26
Show Gist options
  • Select an option

  • Save 23maverick23/d884c6daf4c56892cce232b18d64a9d9 to your computer and use it in GitHub Desktop.

Select an option

Save 23maverick23/d884c6daf4c56892cce232b18d64a9d9 to your computer and use it in GitHub Desktop.
SCLS: Jeopardy Proctor Guide
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Proctor Cheat Sheet — NetSuite AI Jeopardy</title>
<style>
body {
font-family: Georgia, serif;
font-size: 14px;
line-height: 1.6;
color: #111;
background: #fff;
max-width: 900px;
margin: 0 auto;
padding: 2rem 2rem 4rem;
}
h1 {
font-size: 22px;
font-weight: bold;
margin-bottom: 4px;
}
.subtitle {
font-size: 13px;
color: #555;
margin-bottom: 2rem;
padding-bottom: 1rem;
border-bottom: 2px solid #111;
}
h2 {
font-size: 16px;
font-weight: bold;
text-transform: uppercase;
letter-spacing: .06em;
margin: 2.5rem 0 1rem;
padding-bottom: 4px;
border-bottom: 1px solid #ccc;
}
h3 {
font-size: 13px;
font-weight: bold;
text-transform: uppercase;
letter-spacing: .05em;
color: #444;
margin: 1.5rem 0 .5rem;
}
table {
width: 100%;
border-collapse: collapse;
margin-bottom: .5rem;
font-size: 13px;
}
th {
text-align: left;
font-size: 11px;
font-weight: bold;
text-transform: uppercase;
letter-spacing: .06em;
color: #555;
padding: 5px 8px;
border-bottom: 1px solid #ddd;
background: #f7f7f7;
}
th.val-col { width: 60px; }
th.dd-col { width: 28px; text-align: center; }
td {
padding: 7px 8px;
vertical-align: top;
border-bottom: 1px solid #eee;
}
td.val { font-weight: bold; color: #333; white-space: nowrap; }
td.dd { text-align: center; color: #b8860b; font-weight: bold; }
td.clue { color: #222; }
td.ans { color: #1a5c44; font-weight: 600; }
tr:last-child td { border-bottom: none; }
tr:hover td { background: #fafafa; }
.fj-box {
border: 1px solid #ccc;
border-radius: 4px;
padding: 1rem 1.25rem;
margin-top: .75rem;
background: #f9f9f9;
}
.fj-label {
font-size: 11px;
font-weight: bold;
text-transform: uppercase;
letter-spacing: .07em;
color: #777;
margin-bottom: 4px;
}
.fj-clue { margin-bottom: .75rem; }
.fj-answer { color: #1a5c44; font-weight: 600; }
.rules-box {
background: #f5f5f5;
border: 1px solid #ddd;
border-radius: 4px;
padding: 1rem 1.25rem;
font-size: 13px;
margin-bottom: 2rem;
}
.rules-box ul { margin: .4rem 0 0 1.25rem; padding: 0; }
.rules-box li { margin-bottom: 4px; }
.dd-legend {
font-size: 12px;
color: #777;
margin-bottom: 1.5rem;
}
.dd-star { color: #b8860b; font-weight: bold; }
@media print {
body { padding: 1rem; }
h2 { page-break-before: auto; }
table { page-break-inside: avoid; }
.rules-box { page-break-after: avoid; }
}
</style>
</head>
<body>
<h1>NetSuite AI Jeopardy — Proctor Cheat Sheet</h1>
<div class="subtitle">For facilitator use only &nbsp;·&nbsp; All clues, answers, daily doubles, and Final Jeopardy &nbsp;·&nbsp; Do not distribute to players</div>
<div class="rules-box">
<strong>Quick rules reminder</strong>
<ul>
<li>Teams must respond in the form of a question ("What is…")</li>
<li>Facilitator judges whether the intent is correct — exact wording not required</li>
<li>Daily Double (<span class="dd-star">★</span>): teams enter wagers before the clue is read; wager added or deducted from score based on correct/incorrect</li>
<li>Final Jeopardy: teams enter wagers before the question is revealed; 30 seconds to write an answer; proctor awards or deducts wager</li>
<li>Use the +100/−100 buttons on scorecards for manual point adjustments as needed</li>
</ul>
</div>
<p class="dd-legend"><span class="dd-star">★</span> = Daily Double &nbsp;·&nbsp; Green answers are the target responses — accept reasonable variations</p>
<!-- ══════════════════════════════════════
ROUND 1
══════════════════════════════════════ -->
<h2>Round 1 &nbsp;($200 – $1,000)</h2>
<h3>AI Fundamentals</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$200</td><td class="dd"></td><td class="clue">The property of an AI system where the same input always produces the same output — like a SuiteFlow approval rule.</td><td class="ans">What is deterministic?</td></tr>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">LLM-based systems are described as this — context-aware and flexible, but outputs may vary for the same input.</td><td class="ans">What is non-deterministic (or probabilistic)?</td></tr>
<tr><td class="val">$600</td><td class="dd"></td><td class="clue">The unit of text — roughly one word — that determines the cost and speed of every LLM input and output.</td><td class="ans">What is a token?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">This technique reduces AI hallucination by fetching relevant source data from a knowledge base before the model generates its response.</td><td class="ans">What is RAG — Retrieval-Augmented Generation?</td></tr>
<tr><td class="val">$1,000</td><td class="dd dd-star">★</td><td class="clue">The model's 'working memory' — the maximum amount of text it can read and reason over in a single response.</td><td class="ans">What is a context window?</td></tr>
</tbody>
</table>
<h3>The AI Stack</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$200</td><td class="dd"></td><td class="clue">The bottom layer of the AI stack — ERP, CRM, and HCM systems — described as the source of truth for all AI reasoning.</td><td class="ans">What is the Data &amp; Systems layer?</td></tr>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">The AI stack layer containing SuiteScript, connectors, and MCP services — what the model can call to act on real data.</td><td class="ans">What is the Tools &amp; APIs layer?</td></tr>
<tr><td class="val">$600</td><td class="dd"></td><td class="clue">The layer where LLMs, rules engines, and policies live — responsible for generating output.</td><td class="ans">What is the Reasoning &amp; Models layer?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">The top layer of the AI stack — home to task execution, orchestration, and guardrails — and where agents live.</td><td class="ans">What is the AI Automation layer?</td></tr>
<tr><td class="val">$1,000</td><td class="dd"></td><td class="clue">This is why a company with fragmented data across five disconnected systems has a harder AI problem than one on a unified ERP.</td><td class="ans">What is: agents are only as reliable as the data layer beneath them (no single source of truth)?</td></tr>
</tbody>
</table>
<h3>MCP &amp; Protocols</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$200</td><td class="dd"></td><td class="clue">MCP stands for this — an open, industry-standard interface that lets AI models connect to external tools consistently.</td><td class="ans">What is Model Context Protocol?</td></tr>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">Why MCP is described as 'the plug, not the thinking.'</td><td class="ans">What is: MCP provides the connection interface — reasoning and decision-making happen in the model layer, not in MCP itself?</td></tr>
<tr><td class="val">$600</td><td class="dd"></td><td class="clue">The correct answer when a prospect asks whether adopting NetSuite's MCP-based AI Connector will lock them into Oracle's infrastructure.</td><td class="ans">What is no — MCP is an open standard, not proprietary to Oracle or NetSuite?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">MCP serves as the foundation of this NetSuite capability that enables custom agentic-style workflows today.</td><td class="ans">What is the NetSuite AI Connector?</td></tr>
<tr><td class="val">$1,000</td><td class="dd"></td><td class="clue">This is the honest answer when a customer asks 'Is MCP required to use AI in NetSuite?'</td><td class="ans">What is no — MCP is not required; it enables the AI Connector ecosystem but AI features work without it?</td></tr>
</tbody>
</table>
<h3>Human-in-the-Loop</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$200</td><td class="dd"></td><td class="clue">The HITL category where AI acts fully autonomously — appropriate for low-risk, high-volume tasks like categorizing invoices.</td><td class="ans">What is Automate?</td></tr>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">The HITL pattern where AI generates a recommendation or draft, but no action executes until a human explicitly approves it.</td><td class="ans">What is Suggest and Confirm?</td></tr>
<tr><td class="val">$600</td><td class="dd dd-star">★</td><td class="clue">Three examples of actions that must always remain in the 'Always Human' category — AI must never act alone on these.</td><td class="ans">What are employee data changes, GL overrides, and bank detail updates? (accept any two)</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">In a SuiteAgents workflow, this is why the agent asks for confirmation before sending a collections email — even though it retrieved the data autonomously.</td><td class="ans">What is: sending an email is an action tool that changes state, requiring human approval under HITL?</td></tr>
<tr><td class="val">$1,000</td><td class="dd"></td><td class="clue">The most important reason a finance leader accepts AI in their ERP workflow — it comes down to this design principle.</td><td class="ans">What is deterministic controls around probabilistic reasoning?</td></tr>
</tbody>
</table>
<!-- ══════════════════════════════════════
ROUND 2
══════════════════════════════════════ -->
<h2>Round 2 &nbsp;($400 – $2,000)</h2>
<h3>Agentic Levels</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">The AI maturity level covering Q&amp;A and knowledge retrieval — where SuiteAnswers and Ask Oracle operate.</td><td class="ans">What is Answering (Level 1)?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">The maturity level covering drafting, summarizing, and suggesting — where the GenAI API and Exception Management sit.</td><td class="ans">What is Assisting (Level 2)?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">The maturity level for tool use, writebacks, and transactions — where Ask Oracle and AI Connector operate today.</td><td class="ans">What is Acting (Level 3)?</td></tr>
<tr><td class="val">$1,600</td><td class="dd"></td><td class="clue">The maturity level SuiteAgents targets — orchestrating multi-step workflows across multiple systems.</td><td class="ans">What is Coordinating (Level 4)?</td></tr>
<tr><td class="val">$2,000</td><td class="dd"></td><td class="clue">The highest rung on the AI maturity ladder — where the system learns from outcomes and refines its own behavior — currently described as emerging.</td><td class="ans">What is Optimizing (Level 5)?</td></tr>
</tbody>
</table>
<h3>Workflow Spectrum</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">In the transportation analogy, a train on fixed tracks maps to this type of workflow.</td><td class="ans">What is a traditional deterministic workflow?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">A GPS app that reroutes around traffic but always heads to the same destination — the analogy for this workflow type.</td><td class="ans">What is an AI-enhanced workflow?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">A self-driving car where you only set the goal maps to this — the highest level of the workflow spectrum.</td><td class="ans">What is a fully agentic workflow?</td></tr>
<tr><td class="val">$1,600</td><td class="dd"></td><td class="clue">In an AI-enhanced workflow, this tool still owns the overall sequence even when AI steps are inserted.</td><td class="ans">What is SuiteFlow?</td></tr>
<tr><td class="val">$2,000</td><td class="dd"></td><td class="clue">The single word that best describes what an agent does at runtime that a pre-scripted AI workflow cannot — it makes this.</td><td class="ans">What is a decision (or: dynamically determines the next step)?</td></tr>
</tbody>
</table>
<h3>NetSuite AI Features</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">This NetSuite feature lets users ask business questions in plain English and get answers from their own ERP data.</td><td class="ans">What is Ask Oracle?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">The NetSuite AI capability that automatically surfaces invoice mismatches, duplicate records, and payment anomalies for finance teams.</td><td class="ans">What is AI Exception Management?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">The NetSuite tool that lets developers embed custom generative AI capabilities directly into scripts and workflows.</td><td class="ans">What is the GenAI API?</td></tr>
<tr><td class="val">$1,600</td><td class="dd"></td><td class="clue">This upcoming NetSuite tool will let customers build and configure their own AI agents — no deep coding required.</td><td class="ans">What is Agent Studio?</td></tr>
<tr><td class="val">$2,000</td><td class="dd"></td><td class="clue">The NetSuite capability coming in NetSuite Next that orchestrates multi-step agentic workflows — the subject of this entire training.</td><td class="ans">What is SuiteAgents?</td></tr>
</tbody>
</table>
<h3>Positioning &amp; Objections</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">The first question to ask any prospect who opens with 'We want agentic AI' — before demoing anything.</td><td class="ans">What is: 'What specific problem are you trying to solve?'</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">Why a prospect's concern about vendor lock-in is directly answered by confirming MCP's status as this.</td><td class="ans">What is an open, industry-standard protocol (not proprietary to Oracle)?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">The positioning phrase that reassures a CFO or auditor that NetSuite AI doesn't sacrifice control for flexibility.</td><td class="ans">What is: deterministic controls around probabilistic reasoning?</td></tr>
<tr><td class="val">$1,600</td><td class="dd"></td><td class="clue">When a skeptical buyer asks 'How do you stop the AI from making things up in my financial data?', this is the technical answer.</td><td class="ans">What is RAG — the AI retrieves structured data from NetSuite before generating, grounding the output in real records?</td></tr>
<tr><td class="val">$2,000</td><td class="dd dd-star">★</td><td class="clue">The career-limiting risk of overselling AI autonomy before an implementation — what happens post-go-live.</td><td class="ans">What is: the customer loses trust when the system requires more human involvement than was promised?</td></tr>
</tbody>
</table>
<h3>Structured vs Unstructured</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">Why ERP data — rows, schemas, defined fields — is particularly well-suited for AI compared to emails and PDFs.</td><td class="ans">What is: structured data is deterministic, grounded, and auditable — outputs can be traced to source records?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">What AI must do with unstructured inputs like emails or PDFs that it does not need to do with structured ERP data.</td><td class="ans">What is interpret (or: infer meaning using probabilistic reasoning)?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">The type of data where 'the same document can yield different outputs' — a key governance risk to explain to customers.</td><td class="ans">What is unstructured data?</td></tr>
<tr><td class="val">$1,600</td><td class="dd"></td><td class="clue">This is why an AI agent working from a NetSuite invoice record produces a more auditable output than one working from a scanned PDF of the same invoice.</td><td class="ans">What is: the ERP record is structured and traceable; the PDF requires inference to interpret?</td></tr>
<tr><td class="val">$2,000</td><td class="dd"></td><td class="clue">The term for grounding AI responses in real, retrieved data — the technique that makes structured ERP data NetSuite's biggest AI advantage.</td><td class="ans">What is RAG — Retrieval-Augmented Generation?</td></tr>
</tbody>
</table>
<h3>SuiteAgents</h3>
<table>
<thead><tr><th class="val-col">Value</th><th class="dd-col">★</th><th>Clue</th><th>Answer</th></tr></thead>
<tbody>
<tr><td class="val">$400</td><td class="dd"></td><td class="clue">In a SuiteAgents cashflow workflow, these tools retrieve AP balances, AR data, and forecasts — without requiring human confirmation.</td><td class="ans">What are read tools?</td></tr>
<tr><td class="val">$800</td><td class="dd"></td><td class="clue">The type of SuiteAgents tool that sends emails, updates records, or triggers transactions — and requires human approval before executing.</td><td class="ans">What are action tools?</td></tr>
<tr><td class="val">$1,200</td><td class="dd"></td><td class="clue">What SuiteAgents does after autonomously gathering data and synthesizing a recommendation — before taking any consequential action.</td><td class="ans">What is: requests human confirmation (human-in-the-loop gate)?</td></tr>
<tr><td class="val">$1,600</td><td class="dd dd-star">★</td><td class="clue">The honest answer when a customer asks whether SuiteAgents is available to deploy today — accurate and commercially positioned.</td><td class="ans">What is: SuiteAgents is coming in NetSuite Next; AI Connector enables custom agentic flows today; HITL controls are part of the design?</td></tr>
<tr><td class="val">$2,000</td><td class="dd"></td><td class="clue">SuiteAgents sits at Level 4 of the maturity ladder, but this is what separates it from a Level 3 'Acting' capability.</td><td class="ans">What is: SuiteAgents coordinates multi-step workflows across systems — not just a single tool call or transaction?</td></tr>
</tbody>
</table>
<!-- ══════════════════════════════════════
FINAL JEOPARDY
══════════════════════════════════════ -->
<h2>Final Jeopardy</h2>
<div class="fj-box">
<div class="fj-label">Category</div>
<p style="font-weight:bold;margin-bottom:1rem;">NetSuite AI Positioning</p>
<div class="fj-label">Clue (read aloud after wagers are locked)</div>
<p class="fj-clue">A prospect's CFO says: "I'm open to AI in our ERP, but I need to know that my team stays in control of financial decisions and that every AI action is auditable." Give the complete positioning response that addresses both concerns simultaneously — in one sentence.</p>
<div class="fj-label">Correct Answer</div>
<p class="fj-answer">What is: NetSuite uses deterministic controls around probabilistic reasoning — AI provides the contextual intelligence while rules-based guardrails ensure every action is auditable and high-risk decisions always require human approval?</p>
<p style="margin-top:1rem;font-size:12px;color:#777;">Accept any answer that correctly names the principle and addresses both auditability and human control. Exact wording not required.</p>
</div>
</body>
</html>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment