LLM API Selling to the Head of AI Engineering — 60-Min Training

Question

Pulse RevOps · The Machine · Accepted Answer

### Direct Answer

> **LLM API Selling to the Head of AI Engineering** is a 60-minute training for AEs, SEs, and channel sellers running $200K–$5M ACV cycles against incumbents like **Anthropic Claude, OpenAI GPT-5, Google Gemini, AWS Bedrock, Azure OpenAI**. The session teaches sellers to qualify against the **three-buyer reality (Head of AI Engineering, CFO, CISO)**, run a structured discovery on **token economics + frontier benchmarks + compliance**, demo against the customer's actual eval set, and trap-set the multi-year renewal at month 12. Built on **MEDDPICC**, **Force Management's Command of the Message**, and **Andy Paul's "Sell Without Selling Out"** discovery cadence.

---

## Section 1 — Why LLM API Selling Is Different (5 min)

LLM API is not classic SaaS — token economics, frontier benchmarks, and compliance posture all gate every deal. The Head of AI Engineering is the technical buyer; CFO scrutinizes the token bill; CISO governs the compliance and data-handling posture.

Set the frame.

- **Three buyers.** Head of AI Engineering picks; CFO funds; CISO governs.
- **Frontier benchmarks open inbound.** SWE-Bench, GPQA, Chatbot Arena.
- **Caching is the margin lever.** 40–60% input-cost reduction.
- **Compliance gates enterprise.** SOC 2, HIPAA BAA, GDPR DPA, FedRAMP.

End with **Mark Roberge's** rule: *"Sell the eval-set win, not the marketing benchmark."*

---

## Section 2 — The 60-Minute Discovery (15 min)

> 1. **Opening (3 min):** "Walk me through your AI workloads — tokens monthly, current vendor mix, eval set."
> 2. **Token volume baseline (10 min):** "What's your monthly token consumption by use case?"
> 3. **Vendor mix (10 min):** "Are you multi-provider? Anthropic, OpenAI, Google, Llama?"
> 4. **Eval set maturity (10 min):** "Do you have a golden eval set? How many examples?"
> 5. **Compliance posture (8 min):** "SOC 2, HIPAA, GDPR, FedRAMP requirements?"
> 6. **Cache adoption (7 min):** "Are you structuring prompts for cache hit rate above 50%?"
> 7. **Renewal posture (5 min):** "Existing contracts and renewal dates?"

```mermaid
flowchart TD
    A[AE Schedules Discovery] --> B[Send Pre-Brief 24 hrs Prior]
    B --> C{Head of AI + CFO + CISO?}
    C -->|No| D[Reschedule]
    C -->|Yes| E[Token + Vendor Mix 20 min]
    E --> F[Eval + Compliance 18 min]
    F --> G[Cache + Renewal 12 min]
    G --> H[Confirm POC Scope Workshop]
    H --> I[POC Connected Within 7 Days]
```

---

## Section 3 — The POC That Wins (15 min)

**Failure modes to ban.** Sample-eval POCs. No cache discipline. Single-use-case POCs.

**Wins to coach.** **Customer's own eval set ingested.** Walk through **Anthropic and OpenAI POC agendas** — both require golden eval set first. **Side-by-side benchmark scorecard.** **Caching coverage demo.**

End with **Andy Paul's** rule: *"Show the customer their eval set won, not your benchmark expanded."*

---

## Section 4 — Handling the Incumbent (10 min)

**Counter-move 1 — Eval-set wedge.** Ask the Head of AI: *"What's your incumbent's latest score on your golden eval set?"*

**Counter-move 2 — Cache adoption wedge.** Ask the CFO: *"What's your incumbent's cache hit rate? **40–60%** is best-in-class."*

**Counter-move 3 — Frontier benchmark wedge.** *"How does your incumbent compare on SWE-Bench Verified, GPQA Diamond, Chatbot Arena Elo?"*

Show **Force Management's command-of-the-message** rule: *"Displace on customer's eval, not on marketing benchmarks."*

---

## Section 5 — Pricing Conversation (10 min)

**Landmine 1 — Per-token vs. Committed-use.** Committed-use discounts at $1M+ annual spend.

**Landmine 2 — Multi-year discount math.** Three-year deals justify 12–18% discount.

**Landmine 3 — The procurement-only meeting.** No procurement-only rule.

```mermaid
flowchart TD
    A[Joint Head of AI + CFO + CISO Buy-In] --> B[Committed-Use Proposal]
    B --> C{Discount Aligned?}
    C -->|No| D[Reset to Retention Math]
    C -->|Yes| E[MSA Drafted]
    E --> F{Procurement Solo Meeting?}
    F -->|Yes| G[Refuse]
    F -->|No| H[Joint Negotiation]
    G --> H
    H --> I[Onboarding Within 7 Days]
    I --> J[First Eval Run Month 1]
    J --> K[Quarterly Benchmark Review]
```

---

## Section 6 — The Trap-Set for Renewal at Month 12 (5 min)

**Trap-set 1 — Cache adoption above 50% within 6 months.**

**Trap-set 2 — Eval-set win on weekly bake-off.**

**Trap-set 3 — Compliance certification verified at bind.**

**Trap-set 4 — Joint engineering dashboard in QBR.**

Close with **Jeb Blount's** rule: *"The renewal is sold on day one."*

---

## FAQ

**Anthropic or OpenAI default?** Anthropic for coding + safety; OpenAI for reasoning + multimodal.

**Self-hosted Llama competitive?** At 5B+ tokens/month with GPU capacity, yes.

**Eval set size target?** 150–500 examples.

**Cache adoption target?** 50%+.

**Multi-year discount target?** 15–25%.

## Sources

- Anthropic — Customer Outcomes Reference
- OpenAI — Enterprise API Documentation
- Google — Gemini API Reference
- AWS — B

LLM API Selling to the Head of AI Engineering — 60-Min Training

Direct Answer

Section 1 — Why LLM API Selling Is Different (5 min)

Section 2 — The 60-Minute Discovery (15 min)

Section 3 — The POC That Wins (15 min)

Section 4 — Handling the Incumbent (10 min)

Section 5 — Pricing Conversation (10 min)

Section 6 — The Trap-Set for Renewal at Month 12 (5 min)

FAQ

Sources

What does the score mean?