Agents are still a misaligned concept in AI. While this article offers a lot in orchestration, memory (only mentioned once in the post) and governance are not r

Agents are still a misaligned concept in AI. While this article offers a lot in orchestration, memory (only mentioned once in the post) and governance are not really mentioned. The latter is important to increase reliability -- something Ilya Sutskever mentioned to be important as agents can be less deterministic in their responses. Interestingly, "agency" i.e., the ability of the agent to make own decisions is not mentioned once. I work on CAAs and document my journey on my substack (https://jdsmerau.substack.com https://jdsmerau.substack.com)