In this episode we unpack Longrider and the move to long-form AI writing: why current models stall at a few thousand words, how training data length matters, and how AgentWrite's planning-plus-writing pipeline pushes outputs to 10,000+ words. We discuss LongWriter data sets, LongBenchWrite benchmarks, ablation tests, and what this means for the future of AI-assisted storytelling and work documents—along with the limits and open questions that still loom.

Note:  This podcast was AI-generated, and sometimes AI can make mistakes.  Please double-check any critical information.

Sponsored by Embersilk LLC