Your AI Agent Keeps Making the Same Mistakes - Sparsh Jain, CentralAgent AI

Sparsh Jain, Head of AI at CentralAgent AI, argues that giving AI agents more memory isn't the same as helping them learn. After his team worked with the same agent for two years and it still repeated the same mistakes, they rebuilt memory as an experiential system: forming notes from new experiences, updating or deactivating old ones, keeping inactive notes as a record of how a judgment was reached, and forgetting what no longer matters, with a regular cleanup loop that works like a good night's sleep. He shares the rules they learned building it and why agents that learn from every teammate keep compounding over time. Sparsh Jain https://www.linkedin.com/in/jainn-sparsh/ Join us in SF for the second annual AIE CODE summit: https://aie.codes our last flagship summit of 2026. Applications live now!