Google Scientist on Long-Horizon Agents: Balancing Context vs. Scaffolding

AnneliesGamble · x · 2026-08-26

Google DeepMind Distinguished Scientist Prateek Jain discusses two high-level strategies for enabling AI agents to tackle complex, long-horizon tasks: expanding the context window to hold more of the problem in mind, or building external scaffolding (retrieval, memory, planning, tools, sub-agents) to reduce the burden on the model. The interview explores finding the right balance between these approaches.

Related event: DeepMind Scientist on Long-Horizon Agents: Context vs Scaffolding(2 posts)→

Original post →

More from coding & agent

coding & agent channel →