P / Procurement

P / Procurement, Public Insights|公开洞察

P-022|当AI进入采购决策:企业需要重新设计什么?|When AI Enters Procurement Decision-Making

Alice PLSC Insights article header
P-022 | PROCUREMENT & AI GOVERNANCE | OCTOBER 2026

当AI进入采购决策:企业需要重新设计什么?

从Gartner Procurement Conference 2026看采购组织设计与判断治理

新闻观察|2026年9月15—16日
Gartner在美国圣地亚哥举行Procurement Conference 2026。公开议程将AI应用、采购组织运营模式、人才配置与治理问题带入讨论。其中包括“Empowering Procurement to Lead in the Augmented Era”“Preparing CPOs for AI’s Impact on Procurement Talent”以及“Capability-Driven AI Sourcing: Ensuring Readiness, Governance and Safe Adoption”等议题。
本文依据Gartner公开会议公告与议程展开独立分析,不代表Gartner演讲者的完整观点。

当AI开始参与供应市场研究、报价比较、供应商评估与采购流程,企业面临的变化不只是“能否用AI提高效率”,而是:谁完成工作?谁形成判断?谁有权确认或否决?现有流程与组织治理是否仍然适用?

一、AI改变的不只是工具,而是采购工作的组织条件

第一层变化是任务:资料收集、分析与初步建议可以更快完成。第二层变化是角色:采购人员、品类经理、业务部门、技术团队与AI系统之间的分工可能改变。第三层变化是判断治理:AI提出的建议能否进入决策、何时需要人工确认、出现冲突时由谁负责,取决于组织的权限、流程和控制条件。

这三层不能混为一谈。AI生成更好的建议,并不意味着企业已经建立让这些建议发挥作用的治理机制。

二、一个采购情境:技术能力不等于组织允许

假设AI分析发现,某供应商单价较高,但综合物流、质量与使用成本后,总成本更低。采购人员可能接受这一判断;但如果绩效指标仍只考核采购单价,或审批流程不允许采用综合成本依据,这项判断就可能无法进入正式决策。

AI提供了新的判断输入,不代表组织已经给予它适当的判断权限。 技术部署与治理改变是两项不同的管理工作。

三、独立研究的时间线:不同路径,相近的问题

Alice PLSC Insights关于采购判断与组织治理的研究始于2025年初。我们从长期采购实践出发,逐步形成《采购认知金字塔——采购判断治理模型》,研究采购判断在组织中如何发生,以及权限、流程、组织结构和其他治理条件如何影响判断的形成与实现。

2026年9月的Gartner采购大会公开讨论AI、组织设计、人员配置和治理议题。我们注意到,这些议题与自身研究存在交汇,但研究起点、语言与范围并不完全相同。我们不知道Gartner相关研究始于何时,也不据此主张谁先发现了这些问题。

我们并不是因为大型研究机构开始公开讨论这些问题,才开始思考它们。早在这次大会召开之前,我们已经形成了一套以采购判断为研究对象、解释其在组织中如何发生以及受到何种治理条件影响的独立分析方式。

机构规模并不决定研究问题是否值得提出。更重要的是:分析能否解释现实,方法能否被企业使用,结论能否经受后续实践检验。

四、给企业的一份五步检查路径

ALICE PLSC INSIGHTS · PRACTICAL KNOWLEDGE · P-022-A

AI进入采购后的组织治理检查路径|AI-enabled Procurement Governance Review

1. Task|工作任务

列出AI实际参与的采购活动:研究、分析、推荐、执行或监测;区分辅助与自动化。
2. Role|角色分工

明确采购、业务、技术、财务及AI系统分别承担什么工作;识别岗位和能力缺口。
3. Authority|判断权限

标出谁提出建议、谁验证依据、谁批准或否决、何时必须升级处理。
4. Governance|治理条件

检查现有流程、绩效指标、数据规则、控制措施与责任机制是否支持合理判断。
5. Review|持续复核

用真实业务结果检查AI建议的质量、人工干预和组织阻力,并定期调整分工与规则。

怎么用:选择一项已经引入或计划引入AI的采购流程,召集采购负责人、业务代表与相关职能,用五个问题逐项讨论。每一步记录“现状—问题—责任人—下一步行动”,形成一页内部治理检查表。没有部署AI的企业,也可以用相同路径检查现有采购判断机制。

框架状态:基于 Alice PLS Team Insights 采购实践与本次议题形成的实务检查工具(V1.0);并非Gartner框架,也不宣称已完成跨行业实证验证。

五、真正需要重新设计的是什么?

AI可能使采购组织更快获得信息、形成方案和识别风险,但不能自动消除组织中的权限冲突、流程限制与责任空白。企业需要审视的,不仅是采购岗位是否减少或增加,更是判断从产生到被允许、被约束、被采纳和执行的完整过程。

当企业准备引入AI时,是否也准备好重新审视采购判断在组织中如何发生?

Further Exploration|延伸学习

何芳著《采购认知金字塔——采购判断治理模型》正式封面

《采购认知金字塔——采购判断治理模型》|何芳(Fang He)著

本书研究采购判断在组织中如何发生,以及权限、流程、结构和治理条件如何影响判断。无论企业是否已经部署AI,都可以借此进一步梳理采购判断与治理机制。

Procurement Judgment Governance Workshop|采购判断治理工作坊
两天核心工作坊;另设一天可选的进阶实践,面向希望进一步开展组织分析与实际应用的参与者。

作者:何芳(Fang He)|采购决策顾问 · 采购判断专家

参考来源|Sources

Gartner: Conference announcement, June 3, 2026
Gartner: 2026 conference session archive
Gartner: Capability-Driven AI Sourcing session
Gartner: AI and Procurement Talent session

RESEARCH TAGS|AI Governance · Procurement Judgment · Organization Design · Decision Authority · Procurement Talent · Gartner Procurement Conference 2026


ENGLISH VERSION
P-022 | PROCUREMENT & AI GOVERNANCE | OCTOBER 2026

When AI Enters Procurement Decision-Making: What Must Organizations Redesign?

A Perspective on Organizational Design and Judgment Governance Following Gartner Procurement Conference 2026

Conference context | September 15–16, 2026
Gartner held its Procurement Conference 2026 in San Diego, California. Its public agenda addressed AI adoption, procurement operating models, talent and governance. Sessions included “Empowering Procurement to Lead in the Augmented Era,” “Preparing CPOs for AI’s Impact on Procurement Talent,” and “Capability-Driven AI Sourcing: Ensuring Readiness, Governance and Safe Adoption.”
This article draws on Gartner’s public conference announcement and session descriptions. The analysis that follows is independent and should not be attributed to Gartner speakers.

As AI becomes involved in supply-market research, bid comparison, supplier evaluation and procurement workflows, the question is no longer simply whether AI can improve efficiency. It is also: Who performs the work? Who forms a judgment? Who may approve or reject it? Do existing processes and governance arrangements remain fit for purpose?

1. AI Changes More Than Tools: It Changes Organizational Conditions

The first change concerns tasks: information gathering, analysis and preliminary recommendations may become faster. The second concerns roles: the division of work among buyers, category managers, business functions, technical teams and AI systems may shift. The third concerns judgment governance: whether AI-supported recommendations enter a decision, when human validation is required, and who remains accountable depend on organizational authority, processes and controls.

These are distinct changes. Better AI-generated recommendations do not mean that an organization has established the conditions necessary for those recommendations to influence decisions.

2. A Procurement Scenario: Technical Capability Is Not Organizational Permission

Suppose an AI analysis identifies a supplier with a higher unit price but a lower total cost after logistics, quality and usage costs are considered. A procurement professional may accept the analysis. Yet if performance metrics still reward unit-price reductions alone, or approval rules do not accommodate a total-cost rationale, the judgment may never become a formal decision.

AI may provide a new input to judgment without receiving the organizational permission needed for that input to matter. Technology deployment and governance redesign are different management tasks.

3. An Independent Research Timeline: Different Paths, Related Questions

Alice PLSC Insights began researching procurement judgment and organizational governance in early 2025. Drawing on years of procurement practice, this work developed into The Procurement Cognition Pyramid—Procurement Judgment Governance Model, which examines how procurement judgments arise within organizations and how authority, processes, structures and other governance conditions affect their formation and realization.

At its September 2026 conference, Gartner publicly addressed AI, organizational design, talent and governance. We see meaningful intersections with our own research, while recognizing that the starting points, terminology and scope are not identical. We do not know when Gartner began its related research and make no claim to having identified these issues before Gartner.

We did not begin asking these questions because a major research organization started discussing them publicly. Before this conference took place, we had already developed an independent analytical approach centered on procurement judgment: how it arises within organizations and how governance conditions shape what happens to it.

The size of an institution does not determine whether a research question deserves attention. What matters is whether the analysis explains reality, whether organizations can apply the method, and whether its conclusions withstand further examination.

4. A Five-Step Governance Review That Organizations Can Use

ALICE PLSC INSIGHTS · PRACTICAL KNOWLEDGE · P-022-A

AI-enabled Procurement Governance Review

1. Task

Identify where AI participates in procurement: research, analysis, recommendations, execution or monitoring. Distinguish assistance from automation.
2. Role

Clarify the responsibilities of procurement, business functions, technology, finance and AI systems. Identify capability gaps.
3. Authority

Specify who proposes, validates, approves or rejects a judgment, and when escalation is required.
4. Governance

Test whether workflows, performance measures, data rules, controls and accountability arrangements support sound judgment.
5. Review

Examine actual outcomes, AI recommendation quality, human interventions and organizational friction; revise roles and rules accordingly.

How to apply it: Select one procurement workflow using, or preparing to use, AI. Bring together the procurement owner, business stakeholders and relevant functions. For each step, document the current state, the gap, the accountable owner and the next action on a one-page governance review. Organizations not yet using AI can apply the same questions to existing procurement judgments.

Framework status: Practice-oriented review tool (V1.0), derived from Alice PLS Team Insights procurement practice and the issues examined here. It is not a Gartner framework and has not been presented as a cross-industry empirically validated model.

5. What Really Needs to Be Redesigned?

AI may help procurement organizations obtain information, develop options and identify risks more quickly. It does not automatically remove conflicting authority, procedural constraints or gaps in accountability. The real question is not only whether roles or headcount change, but how a judgment moves from being formed to being permitted, constrained, adopted and implemented.

When an organization prepares to introduce AI, is it also prepared to reconsider how procurement judgment actually happens?

Further Exploration

何芳著《采购认知金字塔——采购判断治理模型》正式封面

The Procurement Cognition Pyramid—Procurement Judgment Governance Model | Fang He

This book examines how procurement judgment occurs within organizations and how authority, processes, structures and governance conditions shape it. Its perspective is relevant whether or not an organization has deployed AI.

Procurement Judgment Governance Workshop
A two-day core workshop, with an optional third day for participants who wish to move further into organizational analysis and practical application.

Author: Fang He | Procurement Decision Advisor · Judgment Specialist

Sources

Gartner: Conference announcement, June 3, 2026
Gartner: 2026 conference session archive
Gartner: Capability-Driven AI Sourcing session
Gartner: AI and Procurement Talent session

RESEARCH TAGS|AI Governance · Procurement Judgment · Organization Design · Decision Authority · Procurement Talent · Gartner Procurement Conference 2026

Alice PLSC Insights footer
P / Procurement, Public Insights|公开洞察

P-016|当 McKinsey 开始谈 Procurement Judgment

Alice PLSC Insights

Public Insight|公开洞察 本篇为 Alice PLSC Insights 公开文章。未经授权请勿全文复制或用于商业用途;引用或摘录请注明来源。

ALICE PLSC INSIGHTS · 2026.10.07 · P / PROCUREMENT

AI进入采购之后,什么才是企业真正应该留下来的能力?

2026年9月23日,McKinsey发表了一篇很值得采购从业者阅读的文章:Rewiring procurement in resources: Making judgment an enduring asset。

真正让我停下来的,不是“AI进入采购”这件事本身,而是标题里的这句话:

01|采购真正稀缺的,可能不是数据

在资源型行业,采购团队经常经历一种 Feast–Famine Cycle:市场上行时迅速扩张,市场下行时人员缩减。

问题是,当有经验的 Category Manager 离开以后,真正消失的往往不只是一个人,而是这个人多年积累下来的供应市场理解、谈判经验、历史决策背景,以及那些很难写进流程文件里的判断。

下一轮周期重新开始,企业又要重新学习。

McKinsey提出的一个重要方向,就是利用AI把这些分散在个人身上的 Category Expertise 更系统地捕捉、组织和复用。

这意味着AI在采购中的价值,可能不仅仅是自动化几个任务。

02|从 Data 到 Intelligence Spine

McKinsey把支撑这些采购工作流的数据与知识基础称为 Intelligence Spine。

它连接的不只是合同和发票,还包括 Vendor Master、Rate Cards、Asset Criticality、Work Orders,以及历史采购策略和专家经验。

这里有一个很值得采购团队注意的变化:

McKinsey给出的一个案例中,一家化工企业把历史采购策略、专家经验和分析能力放进同一个平台,在下一轮招标之前实现了大约 14%的海运费率改善,同时也加快了新 Category Manager 的上手速度。

这已经不只是效率问题。

它开始触及一个更深的问题:

03|但是,知识留下来,不等于判断留下来

这是我读这篇文章时最想继续往前走的一步。

AI可以帮助企业保存历史策略,可以整理合同,可以持续监测市场,可以找到价格变化,也可以根据已有知识提出多个采购方案。

但真正进入决策的时候,采购仍然需要回答:

  • 现在应该重新谈判,还是继续执行合同?
  • 什么时候应该寻找第二个供应商(Second Source)?
  • 价格更低的供应商,是否真的值得切换?
  • 供应连续性和短期成本发生冲突时,应该怎样取舍?
  • AI给出的几个方案都“有道理”时,到底应该选择哪一个?

04|AI越强,采购判断可能反而越重要

过去,采购人员面对的问题常常是信息不够。

未来的问题可能恰恰相反:AI会快速生成更多信息、更多分析、更多选项。

于是采购工作的核心链条可能逐渐变成:

当选择越来越多的时候,真正稀缺的能力不一定是“找到答案”,而是判断:

05|能力之外,还有 Decision Rights

McKinsey在文章里还提出了一个非常重要的设计问题:Decision Rights。

不同采购决策,并不应该给AI相同的权限。

这意味着企业面对的已经不只是一个技术问题:

这两件事情不能混在一起。

06|Workflow is the unit of value

McKinsey还有一句很实用的话:

也就是说,企业不应该因为“要做AI”而做AI,而应该从真正能够创造价值的采购工作流开始。

文章重点讨论了三个方向:Category Management、Continuous Sourcing 和 Leakage Control。

例如在 Leakage Control 中,AI可以把 Contract → PO → Invoice → Actual Delivery 连起来,持续寻找合同执行与实际支出之间的偏差。

McKinsey提到,一家能源企业在超过40亿美元的供应商支出中识别出超过1.2%的支出泄漏机会,潜在年度价值约5500万美元;另一家公用事业企业分析超过100万张发票和600多份合同,准确率超过99%,形成超过1000万美元的年度运行价值。

这也是一个很好的提醒:

07|Making Judgment an Enduring Asset

我很喜欢McKinsey这次使用的这个表达。

因为它把采购能力讨论从“一个采购人员会什么”,往前推了一步:

一个组织究竟能够留下什么?

数据可以留下。

流程可以留下。

合同可以留下。

知识正在越来越容易留下。

下一步真正值得研究的是:

如果可以,那么AI对采购真正深远的影响,也许并不是减少了多少采购人员,而是第一次让企业有机会把原本高度依赖个人的经验与判断,逐渐变成一种可以积累、传递、挑战和持续改善的组织能力。


Further Reading|进一步阅读

《采购认知金字塔——采购判断治理模型》何芳著

《采购认知金字塔——采购判断治理模型》
Procurement Cognition Pyramid — Procurement Judgment Governance Model
何芳 / Fang He

如果McKinsey这篇文章讨论的是如何借助AI保存和放大 Category Expertise,那么《采购认知金字塔——采购判断治理模型》进一步追问的是另一个问题:

  • 采购判断究竟是怎么形成的?
  • 什么因素允许、约束或抑制判断的形成与表达?
  • 个人判断与组织决策之间,究竟是什么关系?

两者并不是同一个研究问题。一个更接近AI时代的采购知识与类别经验如何成为组织资产;另一个关注采购判断本身,以及判断如何在组织中形成、运行和被治理。

SEARCH TAGS|Procurement Judgment · AI in Procurement · Category Management · Decision Rights · Intelligence Spine · Procurement Governance

Sources|资料来源

• McKinsey & Company, Rewiring procurement in resources: Making judgment an enduring asset, September 23, 2026.

说明:文中关于采购判断、AI与采购治理关系的讨论,属于 Alice PLSC Insights 基于公开信息与采购实践提出的分析,不是上述来源的原文结论。

Alice PLSC Insights 版权与作者信息

欢迎在文章下方留下不同意见、补充与期待。具有信息价值的评论,经审核后可精选进入 Member Voice。

滚动至顶部