Inbound Logistics | July 2025

How Agentic AI Can Reshape Procurement By VICTOR KUSHCH , Co-founder & CTO, Fairmarkit Procurement

decision-making, and supplier relationship management. Agentic AI isn’t just about

time, and optimize outcomes with minimal oversight. An AI agent can analyze supplier databases, vet candidates against compliance standards, negotiate pricing, and nalize contracts. As a result, the procurement function operates as a business accelerator rather than a bottleneck. The rise of agentic AI won’t replace procurement professionals; it will redene their role. As AI agents take over execution-heavy tasks, procurement teams will shift their focus to strategic oversight, AI-driven

Agentic AI can elevate the

procurement team’s operating model and strategic inuence. Unlike traditional

automating tasks—it’s about embedding AI-driven reasoning into procurement. One of the most valuable skill sets won’t be running sourcing events manually but designing and optimizing AI agent strategies to drive business impact. Agentic AI : The use of intelligent agents to help process and simplify complex data. Capable of adaptation and autonomous problem solving.

AI tools that require human intervention at every step, agentic AI introduces AI-driven agents that operate autonomously, using advanced reasoning and decision- making to manage sourcing tasks end-to-end. These agents don’t just assist in procurement—they execute strategies, adapt to new data in real

Start with the Fundamentals to Unlock Value By SHAWN VO , President and CTO, Denim

inconsistent paperwork, reducing repetitive approvals, or helping teams manage workows that don’t scale well with headcount. A good starting point: look for tasks that are high-volume, repeatable, and follow a general pattern—but still need occasional human judgment. That’s where AI tends to shine. 2. Place Guardrails on AI AI systems in freight often touch sensitive data—everything from banking details and invoices to customer rates and load data. That’s why security and oversight aren’t features to layer on later—they have to be part of the system from day one. If you’re building or buying AI into your business, it helps to ask: What kind of data will this system use? Can we explain how decisions are made? How will we track its performance, accuracy, and fairness over time? One way to think about AI safety is the “Swiss cheese” model. A single layer of protection might have holes, but stack enough layers—training, monitoring, human review, audit logs—and those gaps start to disappear. Each layer strengthens the system. In practice, this might include: • Training AI models with relevant, representative data • Monitoring performance as things change • Keeping a human involved in higher-stakes workows • Logging decisions and outcomes for transparency

AI could add $1.2 trillion in value to supply chains globally, a McKinsey report estimates. But capturing that value depends on how—

and where it’s applied. Across the freight industry, AI is already accelerating everyday work: speeding up quoting, interpreting documents, and enabling faster, more accurate payments. But it’s not one-size-ts-all.

Choosing the Right AI Tools for the Job Not all AI is created equal. Some systems follow rigid rules, others analyze past data to predict

what’s next, and some make sense of messy, unstructured information. Think of it like this: • Rules-based logic handles consistent, structured workows • Predictive models support planning with historical trends • Language models and AI agents spot anomalies in documents and freeform data

Here are three core principles for using AI effectively in freight operations:

1. Start with Usefulness—Not Possibility The best way to evaluate AI isn’t by asking what it could do. It’s by asking what’s slowing your team down today. The most effective use cases aren’t trying to do everything. They’re solving well-scoped problems—like interpreting

Whether it’s reviewing a document or verifying a payment, it should always be clear what the AI is doing—and why.

➥ Continues on page 116

July 2025 • Inbound Logistics 115

Powered by