Did you miss the recent webinar “Agents of Change: How Agentic AI Is Rewriting the Source-to-Pay Playbook,” featuring Andrew Bartolini, founder and chief research officer for Ardent Partners, and Paul Blake, senior director, engagement for GEP?
The webcast provided a practical look at how autonomous agents are transforming source-to-pay operations, including how AI can enhance sourcing, contracting, compliance and payment with greater context, initiative and adaptability.
Today is Part Two of a two-part article series that brings forth the key points from the webcast, with a link to the event.
AI in Procurement: Moving from Rules to Reasoning
For years, automation in procurement has been driven by rules. The dominant model has been a world of “if this, then that,” representing a framework where logic dictates outcomes, and processes follow clearly defined paths. This rules-based automation has brought remarkable efficiency, handling tasks such as routing purchase requests, matching invoices, or recommending sourcing strategies based on specific data triggers. It excels in speed and consistency, but it is also fragile. The moment a variable changes (for example, a supplier updates their invoice format or a data source shifts), the automation may fail. These systems cannot adapt or learn; they must be manually rebuilt and closely monitored for exceptions.
What makes artificial intelligence different, and what makes this moment so important, is that AI introduces adaptability. Instead of operating within strict boundaries, AI learns, reasons, and interacts in ways that resemble human judgment, adding a dynamic new layer to procurement operations.
AI Challenges Remain
Modern AI, particularly in its generative and agentic forms, represents a significant departure from the static automation of the past. It uses machine learning, natural language processing, and probabilistic reasoning to understand context, interpret intent, and make predictions based on complex data. This allows AI systems to summarize vast amounts of information, identify meaning, and generate insights that traditional automation could never deliver. It is not simply faster automation; it is automation that evolves.
The shift from hard-coded logic to context-based reasoning expands what procurement teams can do with data. Instead of following rigid instructions, AI can process unstructured inputs and respond intelligently to new scenarios. This transformation moves automation from the realm of mechanical execution into the space of decision support and adaptive strategy.
However, this advancement comes with serious limitations that every organization must consider. AI systems, for all their sophistication, are capable of producing convincing but inaccurate results. They can generate outputs that appear thoughtful and precise, yet are based on flawed assumptions or incomplete data. The problem lies in the “appearance” of understanding. AI does not truly comprehend context; it mimics comprehension based on patterns it has seen before. It can amplify bias, misinterpret intent, and make errors that humans might never consider. When an AI system presents a decision in a confident, human-like way, it can be difficult to tell when it is wrong. That illusion of intelligence can create risk if not checked by human oversight. As one expert put it, allowing AI to independently run a procurement process today would be like asking a toddler to manage your company — full of energy and potential, but lacking judgment and maturity.
Combining Human Judgment and Machine Speed
Despite these challenges, AI is not a passing trend. It is as transformative as the internet was in the 1990s, and it will reshape procurement in profound ways. The key is in how it is implemented. Procurement leaders must approach AI adoption deliberately by testing, auditing, and scaling gradually. Governance and transparency are essential. Teams need to understand not just what AI can do, but how it arrives at its conclusions. Early pilots and proof-of-concept projects can help organizations build confidence in AI’s capabilities while identifying its limits. In this way, companies can walk before they run, expanding AI’s role responsibly while maintaining trust in the system’s outputs.
Human intelligence remains an irreplaceable part of this evolution. The critical thinking, ethical reasoning, and strategic insight that procurement professionals bring are essential to ensuring AI creates value rather than risk. When two AI agents negotiate with one another, the potential for creative, mutually beneficial outcomes diminishes. The real advantage lies in combining the strengths of human judgment with the speed and precision of machine analysis. Skills such as negotiation, adaptability, and ethical oversight will define the next generation of procurement talent.
As organizations integrate AI into their processes, they must invest in these uniquely human capabilities. Machines are excellent at pattern recognition and data analysis, but they lack empathy, intuition, and moral awareness. These are the qualities that will preserve balance and purpose in an AI-driven enterprise.
A Layered Approach for Balance and Efficiency
There are dangers in assuming that AI’s apparent intelligence is equivalent to real understanding. Some companies will overestimate the technology’s readiness and delegate too much control too soon. The result may be poor outcomes, ethical breaches, or reputational damage. However, when used wisely, AI can enhance performance in ways that were once impossible. The most effective approach is augmentation rather than replacement. AI should take on tasks that add little human value, such as form filling, data entry, and routine validation, freeing people to focus on strategy, supplier relationships, and sustainability. In this model, technology amplifies human capability rather than diminishing it.
Determining where to apply AI depends on understanding the nature of spend and supplier relationships. For example, low-risk spot buys or tail spend can be safely managed with high levels of automation or even autonomy. Tactical categories may require a blend of human and AI input, where technology handles data analysis while humans oversee decisions. Strategic categories, which involve critical suppliers or high-value contracts, demand strong human leadership supported by AI insights. This layered approach helps organizations strike the right balance between efficiency and control. It also encourages staff to become comfortable using AI as a partner, not as a replacement.
Adoption Without Overhaul Is Critical
Speed of adoption is another key consideration. Moving too slowly risks falling behind competitors who are already reaping the benefits of AI-powered agility. Yet, moving too fast invites operational and ethical pitfalls. The answer lies in readiness and mindset. Procurement teams should move quickly to educate their people, assess opportunities, and launch pilots, but they should resist the temptation to overhaul entire systems overnight. As with any enterprise technology, the path to ROI runs through thoughtful implementation, training, and trust.
At GEP, for example, hundreds of AI agents are already being developed and tested across the source-to-pay process. These agents are designed not to replace procurement professionals but to support them, handling tasks that cross multiple workflows such as supplier onboarding, risk assessment, ESG monitoring, and spend analysis. The long-term goal is to build an ecosystem of interconnected agents governed by human oversight and guided by clear ethical standards. This kind of integrated, agentic system will take time to mature, but the foundation is already here. The technology is real, the benefits are measurable, and the opportunity is enormous.
AI is a powerful force that must be guided by human intelligence. It can help organizations operate faster, smarter, and more sustainably, but only if paired with critical thinking and ethical judgment. The procurement teams that will thrive in the AI era are those that view technology not as a replacement for people, but as an extension of their capabilities. By automating what machines do best and elevating what humans do uniquely well, organizations can unlock a new level of value creation, resilience, and innovation.
