Notion's AI Lead: Stop Winning on Token Economics, Route 75% of Traffic to Cut Costs
AI Engineer · youtube · 2026-07-24
Sarah Sachs, Notion's AI engineering lead, breaks down how enterprises can escape the "pricing traps" set by frontier model labs. She points out that models are quietly increasing prices every month (via token inflation or rapid deprecation), turning companies into "AI poor" if their business doesn't grow proportionally.
Core Strategy: Win on Product, Not Token Economics
- Your Supplier is Your Competitor: Buying tokens from a lab that also builds first-party products means paying a markup on a markup, with a high risk of lock-in.
- Model Agnostic (AI Switzerland): Notion's Auto Model quietly routes about 75% of traffic based on "cost per capability per second" and swaps providers underneath.
- Tiered Processing: Stop using frontier models for simple tasks (like triaging emails with Opus). Open-weight models should handle the moderate middle, and CPUs should be preferred over GPUs for basic tasks like turning a CSV into a PDF.
The talk concludes with a live demo of Notion Agents orchestrating tasks: scoping a task, tagging teammates, and opening a PR.
More from coding & agent
- Ben Bajarin: Frontier AI is becoming the enterprise cyber-defense stack — BenBajarin · 2026-10-26
- Building AI bots on raw APIs? Devs say models are 'dumb without the harness' — madeofAjala · 2026-10-26
- HarnessMM cuts agent tokens per task by up to 88% while raising success rates — omarsar0 · 2026-10-26
- Microsoft demos intent-classifying agents built with Copilot App and Foundry, deployable across Teams and Outlook — marlene_zw · 2026-10-26
- Looking for a wearable mic with best battery life that feeds recordings to personal agents via MCP — altryne · 2026-10-26
- The identity crisis in software engineering: multiagent systems demand a new problem set — rakyll · 2026-10-26