Scaling to Long Horizons — Ross Taylor & Chengxi Taylor, General Reasoning
General Reasoning is building RL systems to scale agents toward long-horizon tasks
Ross Taylor, former Meta AI reasoning lead on Llama/Galactica and now CEO of London-based RL startup General Reasoning, outlines the algorithmic and compute challenges of scaling AI agents to long-horizon tasks. The talk draws on his firsthand experience with the 2020-2023 language model era and teases upcoming insights on next frontiers. The content is cut off before the substantive technical claims arrive, limiting signal extraction.