Doom and Quinn discuss how AI tools still require subject-matter expertise to produce high-quality outcomes, citing an AI-made Cannes film that cost $500K and required ~3,000-word prompts and film know-how. They explore “harness + model” as the new differentiator, the shift of buying decisions from models to runtime/orchestration, and rising pressure to manage token spend through routing, caps (e.g., per-engineer budgets), deterministic workflows, and multi-model creator/editor review loops. They cover news including Microsoft’s increased focus on its own models, OpenAI’s Codex and prosumer pivot, Anthropic’s credit/token plans, and Salesforce Agentforce surpassing $1.2B with 205% YoY growth. They question usage-based ARR claims, highlight the need for telemetry/observability and forward-deployed engineers, and note opportunities for FinOps-like optimization layers in AI.

00:00 Skills Still Matter

01:28 Omni Demo Setup

02:19 Comic Book Experiment

03:23 AI News Headlines

07:21 AI Film Harness Moat

09:25 Token Spend Reality

12:38 Comic Result Reveal

13:40 GTM AI Workflows

19:55 Multi Model Routing

26:58 Selling Electricity Debate

29:17 Headless Agents Future

32:46 Wrap Up Next Week