面向精益采购团队的代理人工智能：更好的可见性、更快的决策、全面控制

AGENTIC AIFOR LEAN PROCUREMENT TEAMSBETTERVISIBILITYFASTERDECISIONSTOTALCONTROL PROCUREMENT If that immature approach to procurement limits growth duringnormal business cycles, it is even more problematic in today’sbusiness environment, which is defined by uncertainty. For example,2025 has been a year of frequent and unpredictable changes in thetariffs the U.S. imposes on other nations (and the retaliatory tariffsother countries respond with). Often accompanied by geopoliticaltensions and trade restrictions, tariffs can force companies of allsizes to reimagine their suppliers and supply chains. id-market procurement teams face challenges inany macroeconomic environment. Some of thosechallenges reflect the fact that many small andmid-sized firms are both intensely focused on rapid growthand have not yet established the necessary structures andprocesses to make procurement as consistent, efficient, andtransparent as possible.M The challenge can be especially acute when mid-market companieshave various business units implementing their own procurementstrategies and policies. There are real-world implications to afragmented approach to procurement. From wasted staff time andunnecessarily high and unpredictable costs to greater risk of supplydisruptions and poor financial visibility and control, an immatureand ad hoc approach to procurement ultimately makes it far moredifficult for mid-market companies to achieve their growth objectives. This fluid and risky business environment exposes the deficienciesof procurement when it is managed with a patchwork of tools,spreadsheets, and manual work. Beyond the typical risk of errorsand elevated costs, today’s unsettled and often chaotic businessenvironment also highlights the need for greater flexibility inresponding quickly to changing conditions. “Mid-size companies need to move fast,” said Alex Zhong,Senior Director of Product Marketing for GEP, a provider ofprocurement and supply chain software and services. “Limitedresources mean they must operate with greater agility andresilience, making faster decisions and adjusting in real timerather than relying on buffers like excess inventory.” The procurement dilemma mid-market companies face isdriven both by people and process. Like large enterprises,fast-growing mid-market firms need structures, processes, andpolicies to effectively manage their spending. But when theyattempt to implement new procurement tools and procedures,they are often met with resistance—especially when staffers areaccustomed to informal purchase processes. If a new processfeels like extra work, people will avoid it,” Zhong said. “But whenprocurement tools are easy to use and fit naturally into existingworkflows, adoption becomes much easier. “Mid-size companies need to move fast. Limited resourcesmean they must operate with greater agility and resilience,making faster decisions and adjusting in real time rather thanrelying on buffers like excess inventory.” ALEX ZHONG,Senior Director of Product Marketing for GEP THE EMERGENCEOFAGENTIC AI Mid-market firms lacking defined procurement policiesand systems can accelerate their maturity, standardization,and flexibility by implementing agentic AI. Unlike previousgenerations of AI, agentic AI systems are designed to actautonomously, make decisions based on pre-defined guidelines,and orchestrate the complex workflows across multiple systemsthat are so common in procurement. To be clear, agentic AI is not the same as large languagemodels (LLMs) like ChatGPT. While LLMs can answer complexquestions and produce text and graphics to communicateanswers, they depend on people for direction. Agentic AI buildson LLMs but extends beyond them, combining reasoning,orchestration, and action execution to operate autonomouslytoward business objectives. Agentic AI can make sophisticated decisions, providedagents are built and trained to achieve complex outcomes.In procurement, for example, negotiation is a complex andnuanced task that requires synthesizing and analyzing externalmarket data on production costs, competitor pricing, demand,policies such as tariffs, and other relevant information. “Agentsaren’t generalists,” Zhong said. “They’re specialists, built aroundfunctions that can be as sophisticated as the decisions they’redesigned to support.” “Agents aren’t generalists. They’respecialists, built around functions that canbe as sophisticated as the decisions they’redesigned to support.” ALEX ZHONG,Senior Director of Product Marketing for GEP DATA IS THEFOUNDATIONOF AGENTIC AI Agentic AI’s sophistication and capabilities depend oncomprehensive and clean data from across an organization. Thiscan be particularly challenging for mid-market companies. “AIcan’t work without data,” said Raj Aggarwal, Director of ProductMarketing at GEP. “It’s a big problem when you have differentsystems with different types of data in different places, whetherit’s supplier data, ERP data, or other systems.” More specificall