Industry

Harrison Chase 转发观点:企业 AI 落地缺的不是模型,而是知识体系

Why we picked it

LangChain 的 Harrison Chase 转发了 Grant Miller 这套方法:把老员工的流程写成 skills、业务逻辑做成 wiki,再让一个 curator 智能体每晚自动修正文档。买了 Copilot 却觉得没效果的公司,值得照着这套两步框架自查一下知识基建。

LangChain 创始人 Harrison Chase 转发并认同 Grant Miller 的观点:很多公司买了 Copilot 之类的工具,却发现 AI 没有改变团队的工作方式,问题通常不在模型本身,而是智能体不了解公司内部的运作方式。

Body

他给出的解法分两步:第一步把公司最有经验员工的知识显性化,包括把流程写成逐步可执行的 skills、把业务逻辑整理成互相链接的 wiki、明确数据存放在哪里以及如何调用。第二步是引入一个维护智能体:工作智能体在执行过程中标记缺失或错误的信息,然后由一个 curator 智能体每晚处理这些标记并修正源头文档。

这条推文获得了 6 条评论和 14 个点赞。对已经采购了 Copilot 等编程助手但收效不明显的企业来说,这一框架把问题从“选哪个模型”转向“知识基建是否到位”,与当前企业普遍把预算押在模型采购上的做法形成对照。

Original · Harrison Chase

totally agree with this a lot of companies will buy copilot or something and then wonder why AI isnt changing how their team works the knowledge factory is the missing piece Grant Miller @AIGuide_ if AI still isn't changing how your team works, this is probably why... the model usually isn't the problem, your agents just don't know how your company works the fix comes in two parts: first, write down what's in your best people's heads > your processes as step-by-step skills > how the business works in one linked wiki > where your data lives and how to pull it then add an agent that keeps it all up to date > your agents flag anything missing or wrong while they work > a curator agent goes through those flags every night and fixes the source 🔗 View Quoted Tweet📊 6 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include LangChain, Copilot, agents, enterprise knowledge bases.

Source attribution: curated from Harrison Chase (@hwchase17).