2026-09-10
每日一文 · 长文精读
Meta's Recipe for Building Agents as "Organizational Second Brains"
Meta 构建作为“组织第二大脑”的代理的秘诀
作者：Sergio De Simone · InfoQ 原文
摘要：Meta 构建了一个名为“组织第二大脑”的AI代理，通过结构化知识架构和自我改进循环捕获专家知识。系统包括知识系统（使用分类文件）、推理管道（使用“配方”）、评估框架和自我改进循环。关键是将知识存储在版本控制文本文件中而非模型权重，从而无需重新训练即可持续改进。评估显示，它将评估时间从数天缩短到数分钟，且零回归。该方法可推广到安全、金融等领域。
Meta describes how an AI agent can be designed to capture the logic and expertise of domain experts, rather than simply storing documents or retrieving relevant information.
Meta 描述了如何设计一个 AI 代理来捕获领域专家的逻辑和专业知识，而不是简单地存储文档或检索相关信息。
The system, dubbed an "organizational second brain", was built for a specialized compliance domain, but Meta argues the architecture generalizes to areas like security, finance, engineering, and procurement.
该系统被称为“组织第二大脑”，是为专门的合规领域构建的，但 Meta 认为该架构可推广到安全、金融、工程和采购等领域。
We've built an AI agent that acts as a secondary expert for a given domain, making deep specialist knowledge readily available and preserved for anyone in an organization to access, share, and build upon.
我们构建了一个 AI 代理，充当特定领域的辅助专家，使深度的专业知识随时可用并得以保存，供组织中的任何人访问、分享和在此基础上进一步发展。
According to Meta, their approach differs from a traditional domain-specific agent by combining a "structured, auditable knowledge architecture" with a self-improvement loop that "compiles expert feedback into verified, regression-tested updates without model retraining".
据 Meta 称，他们的方法与传统领域特定代理的不同之处在于，将“结构化、可审计的知识架构”与自我改进循环相结合，该循环“将专家反馈编译为经过验证、回归测试的更新，无需模型重新训练”。
The architecture consists of four layers: a knowledge system to consolidate all institutional knowledge; a reasoning pipeline that separates what the agent knows from how it reasons; an evaluation framework to provide automated benchmarks; and the already mentioned self-improvement loop.
该架构由四层组成：整合所有机构知识的知识系统；将代理所知与其推理方式分开的推理管道；提供自动化基准的评估框架；以及前面提到的自我改进循环。
The knowledge system uses expert knowledge distilled from over 200 structured files organized into a strict taxonomy.
知识系统使用从 200 多个结构化文件中提炼出的专家知识，这些文件按严格的分类法组织。
These include position files, capturing authoritative know-how; taxonomy and vocabulary files, acting as a glossary for entity types, activity categories, classification tiers, and other terms used across the organization; routing indexes, mapping inputs to relevant files without relying solely on embedding similarity; and gateway files, defining tests that determine when the agent can safely apply specialized knowledge.
这些包括：position files，捕获权威知识；taxonomy and vocabulary files，作为实体类型、活动类别、分类层级以及组织中使用的其他术语的词汇表；routing indexes，将输入映射到相关文件，而不单独依赖嵌入相似性；以及 gateway files，定义测试以确定代理何时可以安全应用专业知识。
The reasoning layer uses "recipes" to define how the agent should analyze a problem.
推理层使用“配方”来定义代理应如何分析问题。
Recipes are composable and specify what the agent should examine first, which knowledge to load at each step, which decision procedure to follow, and how to determine when the analysis is complete.
配方是可组合的，指定代理应首先检查什么、每一步加载哪些知识、遵循哪个决策程序，以及如何确定分析何时完成。
Recipes also make it possible to determine whether a failure came from missing knowledge or a flawed reasoning procedure.
配方还使得能够确定失败是来自知识缺失还是推理程序有缺陷。
A key feature of Meta's approach is the use of predefined checkpoints in the analysis where human control is required and ambiguous cases are escalated to domain experts.
Meta 方法的一个关键特征是在分析中使用预定义的检查点，在这些检查点需要人工控制，并且模棱两可的情况会升级给领域专家。
When an expert identifies a mistake, the correction is permanently recorded, whether it stems from missing knowledge, a flawed reasoning procedure, or genuine ambiguity among experts.
当专家发现错误时，无论错误源于知识缺失、推理程序有缺陷还是专家之间的真正歧义，修正都会被永久记录。
The key idea that makes this self-improvement flywheel possible is to keep institutional knowledge out of the model's weights and instead store it in version-controlled text files that are accessed by explicit reasoning procedures and rigorously tested:
使这种自我改进飞轮成为可能的关键思想是将机构知识保留在模型权重之外，而是将其存储在版本控制的文本文件中，这些文件由明确的推理过程访问并经过严格测试：
Expert corrections are diagnosed to their root cause, compiled into minimal verified edits, and evaluated against replay and regression tests before they are reviewed and landed.
专家修正会被诊断到根本原因，编译成最小的已验证编辑，并在审查和落地之前根据回放和回归测试进行评估。
Each fix is then folded back into the regression suite, so the gain is permanent.
每个修复随后被合并到回归套件中，因此改进是永久的。
Based on Meta's own assessment, its approach to building organizational experts with agents reduced the time required for individual assessments from days to minutes, with domain experts judging the results useful almost all the time.
根据 Meta 自己的评估，其使用代理构建组织专家的方法将个人评估所需的时间从几天减少到几分钟，领域专家几乎总是认为结果有用。
Furthermore, the system automated knowledge improvements that previously required engineering sprints and experienced zero regressions across improvement cycles.
此外，该系统自动化了以前需要工程冲刺的知识改进，并且在改进周期中经历了零回归。
There is much more to Meta's approach to creating agents capable of serving as organizational experts than can be covered here. Be sure to read the original article for the full details.
Meta 创建能够充当组织专家的代理的方法远不止本文所述。请务必阅读原文了解全部细节。
阅读理解
1. What is the primary difference between Meta's agent and a traditional domain-specific agent?
2. Which type of file in the knowledge system defines tests for safe application of specialized knowledge?
3. What is the key idea that enables the self-improvement flywheel?