And our IT systems just traditionally haven’t been super effective at being able to standardize those orders. We’re able to use AI agents to take hundreds of thousands of man-hours out of the system, where we don’t have to touch orders; they can be standardized, processed in our system, and it nets out in material savings for us. A lot of that work was outsourced already, so that’s not necessarily saving our employees a lot of time, but it is kind of saving the bottom line a bit of time. So when we did a couple of different gut checks on what we think the impact of AI was going to be this year, we saw it in the neighborhood of a million to a million plus man-hours of savings that we can redeploy.
Minimal output tokens. With thousands of configurations to sweep, each evaluation needed to be fast. No essays, no long-form generation.Unambiguous scoring. I couldn’t afford LLM-as-judge pipelines. The answer had to be objectively scored without another model in the loop.Orthogonal cognitive demands. If a configuration improves both tasks simultaneously, it’s structural, not task-specific.The Graveyard of Failed ProbesI didn’t arrive at the right probes immediately; it took months of trial and error, and many dead ends
，推荐阅读新收录的资料获取更多信息
Continue reading...
到了2024年，随着侯毅离开盒马，新零售的变革告一段落。山姆在消费者和零售业同业者中都被架上了“神坛”。而在这一年的演讲中，朱晓静回应了当时外界关于消费升级还是消费降级的争论：“应该说，消费升级、中产的崛起是我们过去这二三十年中国消费市场的主题。随着彭会长刚才提到的‘人均GDP一万美元’这个重要消费拐点的到来，消费者逐步从‘消费升级’的热浪中，回归和进入到‘理性消费’。理性，不是简单的消费降级，而是对‘价值感’更加关注。”