Teenagers are remarkably good at experimenting. They try new things, question established rules, and quickly abandon what doesn’t work. Sourcing and procurement leaders could use some of that attitude toward AI, provided they combine experimentation with clear boundaries.

When it comes to AI in procurement, think like a teenager: Experiment and question the rules, but recognize that you still need boundaries. Our new report, The Forrester Wave™: Supplier Value Management Platforms, Q4 2026, is live! It shows a market racing toward agentic AI, yet for most procurement organizations, the near-term opportunity is augmentation, not autonomy. AI can take on voluminous, manual, and repetitive administration tasks, guide employees, enforce controls, monitor exceptions, and expand procurement capacity.

Tech execs and procurement leaders should regularly confirm that the supplier value management (SVM) suite’s source of truth supports data portability and integration standards and has clear governance. Vet the robustness of point solution specialists’ prebuilt connectors to your primary source of truth, and factor in the time, money, and effort needed to integrate different systems. A hybrid composable procurement architecture may be a better fit: a suite that offers open APIs, prebuilt connectors, and a strong partner ecosystem while relying on a specialist for specific capabilities.

Here are three lessons from the SVM Wave evaluation for approaching AI in procurement.

  1. Experiment With AI, But Give It A Curfew

Don’t start by asking: “How autonomous can we make procurement?” Instead, ask: “Where can an AI agent safely take work off my plate, my team’s plate, and my stakeholders’ plates?”

Give agents boundaries. Let them handle repeatable tasks and guide users, then test how they manage exceptions and when they return decisions to humans. This approach is called controlled augmentation. Buyers should assess whether agents can operate reliably at scale while maintaining decision-making, governance, and exception-handling boundaries. Autonomy should be earned.

  1. Stop Shopping For The Longest Feature List

The SVM market increasingly offers impressive breadth. That makes SVM platform selection deceptively difficult. More features and functionality do not automatically mean more value. Prioritize operational fit. Does the platform fit your workflows, architecture, integration requirements, and complexity? Can it scale without creating a mountain of customization and integration debt?

Our evaluation found that the platforms that create the most value may be those that best fit the enterprise, rather than those with the longest feature list. That distinction becomes even more important with AI. Every new agent sitting on fragmented data and broken processes risks automating complexity rather than removing it.

  1. Measure Outcomes, Not AI Activity

AI demos packed with agents can be seductive. Tech execs and their procurement leaders need a tougher test. Ask instead: “What has changed in your business?” Start with a small number of high-impact use cases. A T-strategy is a useful approach: Start small, dig deep, and establish measurable outcomes before expanding. Learn what works before expanding automation. Early wins can build trust and create a foundation for scaling agentic AI across procurement. The metric should be procurement outcomes, not the volume of AI activity. That may be the most “teenager” lesson of all: Don’t believe something merely because an authority tells you it works. Test it yourself.

Read The Forrester Wave™: Supplier Value Management Platforms, Q4 2026 to explore the evaluation and determine which providers best fit your priorities. Use the Wave as a starting point, then adapt the evaluation to your own requirements.

Special thanks to my colleagues Fred Giron, Eveline Oehrlich, Laura Demarki, and Rachel Birrell, whose support and guidance made this report possible.

Want to discuss what agentic AI means for your procurement technology strategy? Request a guidance session with me.

Share