Rep ramp quiz from call library

❂

Rep ramp quiz from call library

Enablement

Generates a scenario-based onboarding quiz for a ramping rep, built from real winning/losing calls in your library — each question is a constrained micro-moment (the open, the first two discovery questions, the objection turn) with a graded answer key, so new reps drill the exact iCompaas selling motion.

The routing recipe

Job typeCreative & marketingPolicyBalancedConstraintsctx ≥ 128k

routes to

Llama 3.1 Nemotron Ultra 253B v1 nvidia

nvidia:nvidia/llama-3.1-nemotron-ultra-253b-v1

86.3

match score

90.3

task fit

100

cost

53.4

speed

Why this model from benchmarks & capabilities

"Creative & marketing" leans hardest on creative; Llama 3.1 Nemotron Ultra 253B v1 scores 87/100 there — top-1 of 40 chat models.
Strength on this class of work shows up on EQ-Bench, LMArena (human preference), where NVIDIA models are competitive.
Cost: runs on NVIDIA free credits / self-hostable NIM — effectively $0 marginal. Speed: ~45 tok/s (estimated).
Chosen over Qwen3 Coder 480B-A35B Instruct for higher task-fit (90.3 vs 86.7).

Capability the job needs	Weight	Llama 3.1 Nemotron Ultra 253B v1	Catalog rank
creative	1.00	87	#1 of 40
instruction	0.70	96	#1 of 40
knowledge	0.50	91	#7 of 40

Relevant benchmarksEQ-Bench LMArena (human preference) EconomicsCost: runs on NVIDIA free credits / self-hostable NIM — effectively $0 marginal.
Speed: ~45 tok/s (estimated). vs runner-upChosen over Qwen3 Coder 480B-A35B Instruct for higher task-fit (90.3 vs 86.7). (Qwen3 Coder 480B-A35B Instruct, score 86)

Example result

Sample input ▾

Call-library tags for the SMB SOC 2 segment: - WIN (Acme, 45 ppl): prospect opened 'enterprise prospect wants our SOC 2 report'; AE quantified the stalled deal at $180k ARR before demoing; booked dated POC. - WIN (Borealis): AE turned 'we're looking at Drata' into criteria discovery instead of a feature war. - LOSS (Cirrus): AE answered 'how much?' with a number cold, deal stalled; never found the compelling event. - LOSS (Delta): AE demoed 12 features in 8 min; prospect went dark. New rep: week 2, from a non-compliance SaaS background.

RAMP QUIZ — SMB SOC 2 motion (Week 2) — 5 items, pass bar 4/5 green Q1 · The open (easy) [Scenario] Inbound demo. Prospect, a 45-person SaaS, joins and says: "An enterprise prospect is asking for our SOC 2 report and we don't have one." [Prompt] First thing out of your mouth? [Model answer] "That's the #1 reason teams your size call us — let's make sure we can actually unblock that deal. Quick: which prospect, and what's the contract worth to you?" [Why] Acknowledge → reframe to THEIR outcome → start quantifying. Don't pitch yet. [Auto-grade] 🟢 acknowledges + asks a quantifying/qualifying question · 🟡 acknowledges but jumps to product · 🔴 starts demoing features Q2 · First two discovery questions (easy) [Scenario] Same call, 60 seconds in. You have the compelling event but no numbers. [Prompt] Name your first two questions. [Model answer] (1) "What's that enterprise deal worth in ARR?" (2) "What deadline did they give you for the report?" [Why] Dollarize, then date. This is exactly what the Acme WIN did before demoing ($180k, unblocked). [Auto-grade] 🟢 both quantify+time the pain · 🟡 one is a feature question · 🔴 both are product/logistics Q3 · The compelling-event dig (medium) [Scenario] Prospect: "Yeah we'll probably need SOC 2 at some point this year." [Prompt] Turn the vague 'at some point' into a real event — what do you ask? [Model answer] "Totally — is anything forcing the timeline? A specific deal, a renewal, a customer security questionnaire sitting in your inbox?" [Why] No compelling event = Cirrus LOSS. Manufacture clarity by offering the usual triggers. [Auto-grade] 🟢 probes for a dated external trigger · 🟡 asks timeline generically · 🔴 accepts 'someday' and moves on Q4 · The competitor turn (medium-hard) [Scenario] Prospect: "We're also evaluating Drata." [Prompt] Your next sentence? [Model answer] "Makes sense — they're solid. If both of us could get you audit-ready, what would actually make you pick one over the other?" [Why] Borealis WIN: convert competition into decision-criteria discovery; never open with a feature war you can't verify. [Auto-grade] 🟢 elicits decision criteria · 🟡 neutral but no criteria captured · 🔴 trashes Drata / feature-dumps Q5 · The price reflex (hard) [Scenario] Minute 4, no compelling event quantified yet. Prospect: "So how much is this?" [Prompt] Respond. [Model answer] "Happy to — pricing scales with frameworks and your stack, so let me ask two quick things so I quote the right thing, not a random number. What frameworks, and what's driving the timeline?" [Why] Cirrus LOSS came from a cold number with no value anchored. Earn the right to quote. [Auto-grade] 🟢 defers with a reason + re-anchors on value · 🟡 gives a range then digs · 🔴 drops a cold number PASS BAR: 4/5 green to advance to live-call shadowing.

Test it on your own data