Fine-Tuning Platform Selling to the ML Platform Lead — 60-Min Training

Direct Answer

Fine-Tuning Platform Selling to the ML Platform Lead is a 60-minute training for AEs running $50K–$650K ACV cycles against Together AI, Fireworks AI, OpenAI Fine-Tuning, AWS SageMaker, GCP Vertex AI, Modal, Replicate, Hugging Face AutoTrain. Qualify against ML Platform Lead + Head of AI Engineering + CFO, run discovery on base model breadth + training velocity + inference attach + cost.

Built on MEDDPICC.

Section 1 — Why Fine-Tuning Platform Selling Is Different (5 min)

Fine-tuning is bought to specialize behavior + reduce inference cost.

End with Mark Roberge's rule: *"Sell the model-cost-reduction + style-control story."*

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

1. Opening (3 min): "What models are you running today? What gaps prompted fine-tuning?"
2. Base model preferences (10 min): "Llama 4 405B/70B/8B? Mistral? DeepSeek?"
3. Training data volume (10 min): "10K+ labeled examples per task?"
4. Inference-attach posture (10 min): "Self-host or vendor inference?"
5. GPU access (8 min): "Burst training capacity needs?"
6. CFO cost story (7 min): "Per-token economics post-fine-tune?"
7. Renewal posture (5 min): "Existing fine-tuning contracts?"

Section 3 — The POC That Wins (15 min)

Failure modes to ban. No real customer dataset. No inference endpoint attached. No before/after comparison.

Wins to coach. Customer dataset ingested. LoRA fine-tune in 4 hours. Inference endpoint live with side-by-side eval.

End with Andy Paul's rule.

Section 4 — Handling the Incumbent (10 min)

Counter-move 1 — Base model breadth wedge. *"Llama, Mistral, DeepSeek all supported?"*

Counter-move 2 — Training velocity wedge. *"Time-to-first-trained-model?"*

Counter-move 3 — Inference attach wedge. *"Inference endpoint provisioning included?"*

Section 5 — Pricing Conversation (10 min)

Landmine 1 — Per-token vs. Per-job. Both required.

Landmine 2 — Multi-year discount. 12–18% for 3-year.

Landmine 3 — No procurement-only meetings.

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

Trap-set 1 — 5+ fine-tunes per customer per quarter.

Trap-set 2 — Inference attach 60%+.

Trap-set 3 — Per-token cost reduction 40%+ vs unfine-tuned baseline.

Trap-set 4 — Joint ML dashboard in QBR.

Close with Jeb Blount's rule.

FAQ

Together AI or Fireworks? Together for breadth; Fireworks for inference speed.

OpenAI fine-tuning competitive? For GPT-5o-mini, yes.

Self-host or managed? Managed under 100M training tokens/month.

LoRA or full FT? LoRA in most cases.

Inference attach rate target? 60%+.

Sources

• Together AI — Fine-Tuning Reference
• Fireworks AI — Fine-Tuning Documentation
• OpenAI — Fine-Tuning API Pricing
• AWS — SageMaker Reference
• GCP — Vertex AI Custom Training Reference
• Hugging Face — AutoTrain Reference
• Modal — Serverless Reference
• Replicate — Reference
• Force Management — MEDDPICC
• Mark Roberge — Sales Acceleration Formula