
We’ve cut the cost of an agent conversation 75% this year, even as the agent got far more powerful, with more context and more background curation. Our CTO Caitlin Colgrove on how we did it, and why we optimize for intelligence per dollar, not tokens.















