Cut 25-33% of agent input token costs by giving your bot a second brain in Notion

bfrench · x · 2026-11-18

A practical workflow for managing AI assistant usage limits: the author built a "second brain" for his bot in Notion, letting him see all ongoing work at a glance, direct the bot to fetch only specific data, and better control context compaction. The key economics: models charge for both input and output tokens, and input tokens are often 25-33% of total cost — so compress both sides to stay within usage allotments.

Original post →

More from coding & agent

coding & agent channel →