Agentic AI’s Transition from Promise to Proof in Enterprise Procurement

For years, the conversation surrounding AI in procurement has been dominated by its potential. It offers the promise of future efficiencies, productivity and better business outcomes that always seemed one pilot program away. Now, with the rise of agentic AI, we’re seeing a welcome shift away from speculation about the promise to showing actual proof of ROI in this transformative technology.

Indeed, several industry reports say 2026 should be the year enterprises “get real” with AI by starting to measure value. These align with what we’re hearing from the G2000 procurement leaders we work with day to day, who are publicly reporting the many ways it’s contributing to their bottom lines.

Tesco and HP show what can be done with AI

One company that got ahead of the curve was our customer Tesco. The UK grocery giant called out its Save to Invest program in their 2024 Annual Report. It credited AI with helping to deliver £1.2bn in savings and enabling it to “offset inflation and create headroom to fund investments.”

More recently, HP reported improved efficiency, lower costs, and gaining control over previously unmanageable tail spend. Steve Dyson, VP and Global Head of Indirect Procurement, said: “The ability to use AI to really streamline price and service comparisons takes a couple of week’s work and moves it into a couple of hours.”

He added, “We can average 10–15% off for competitive buys, generating money off tail spend that really hasn’t been generated before.”

It’s no surprise that agentic AI is excelling in procurement. The function is inherently complex, highly data-driven, and vulnerable to human error, making it an ideal candidate for agentic systems. No procurement team has the capacity to engage every business stakeholder with the speed, ease, and effectiveness demanded by CFOs and CEOs. The complexity is simply too high, volumes are too high, and the time available is too limited.

10x Faster – Half the Resources

Traditional procurement has long been renowned for its friction: lengthy RFP cycles and manual interventions that drag down enterprise speed. AI-enabled sourcing is now delivering a huge boost in throughput with significantly fewer resources.

In a recent webinar, Bristol Myers Squibb explained how it erased 5 months of friction by moving from a 6–9-month RFP cycle to just 27 days. Crucially, they are now executing 10x as many RFPs as during their pre-launch period, while operating with 50% fewer people by bringing work back in-house.

Or take the London School of Economics (LSE). In just three months, the LSE completed 40 tenders, double the institution’s typical annual output.

This addresses the fundamental scaling problem procurement is facing. Cristian Martin, LSE Department of Procurement Services, highlighted in a recent interview: “If I had 4,000 procurement staff, I wouldn’t need AI, but unfortunately, I don’t. I can’t look at every piece of paper that goes through the Finance department, as the scale is just too much.”

Solving the “Inbound Inbox” Problem

While external negotiations often get the headlines, AI agents are also solving a more stealth enterprise speed killer: operational debt caused by internal communication chaos. Huge volumes of inbound requests paralyze many procurement teams. In addition, many arrive late, through ad hoc channels, or without structure.

In the same webinar, Invesco gave a powerful account of this systemic drag. Their procurement inbox was plagued by 12,000 emails per year. However, 80% of those were internal messages from team members trying to move processes along manually. This was a massive drain on their productivity. With agentic AI, they now have a structured gateway to turn their chaotic inbox into streamlined, autonomous workflows.

The Death of the “Black Box” and the Rise of AI Explainability

As companies like these gain more trust and confidence using agentic AI, and roll it out more broadly, the ROI gains stand to multiply. However, as this happens, maintaining trust and AI ‘explainability’ grows in importance.

When you hire someone new, you don’t hand them the keys to the kingdom on day one. You start with close interaction, guiding them through your company’s policies, culture and values, granting incremental autonomy as they prove their reliability. Trust is built when that employee demonstrates that they are following these guidelines and delivering consistent results.

Equally, people want to see and understand a new AI agent’s thinking before putting their trust in them. That’s why as AI agents assume responsibility for billions in spend, the “black box” approach of providing answers without context, is on its way out. In high stakes areas like procurement, it poses unacceptable risks.

Why “No UI with AI” is a Myth

As we have evolved from pilots to production – and measurable value – the next frontier is fully realizing that value through sustained and widespread enterprise adoption. And the extent to which this adoption grows depends largely on user experience (UX).

There is a common industry slogan that “There’s no UI with AI,” suggesting that as systems become smarter, the interface disappears. I believe the opposite: that in reality autonomous systems raise the bar for UX, paving the way for smoother human-to-machine collaboration.

There are two approaches here. The first is obvious: what does that experience look like for the end user? How do humans interact with intelligent systems day to day? The second is more subtle: how do we guide or direct an agent to work for us? That’s a completely new layer of UI that needs to be invented.

The new User Interface

Specifically, a kind of ‘steering layer’ is needed for the UX that extends well beyond the current experience of prompting, where an organisation’s DNA is embedded in agents’ decision-making processes. Far from disappearing, I expect interfaces to increasingly allow humans to steer agents in ways that align with a company’s goals, culture and values.

The Holy Grail here is a new UX that gives users true agency, letting them define the level of automation vs human involvement at every stage in a complex process, while clearly explaining the rationale behind decisions. This gives enterprise users the confidence, control, and the ability to stay in the loop.  It’s also a UX that turns AI agents into trustworthy, transparent partners that are incredibly good at getting their jobs done.

Given the significant progress Agentic AI is making in speed, resolving operational debt, and improving transparency and trust in the user experience, those impressive ROI statements from global enterprises keep coming out. That’s why I firmly believe it’s on track this year to complete its transition from ‘promise’ to ‘proof’ in enterprise procurement.

Globality’s white paper “Reimagining Sourcing with AI: A Practical Roadmap to Building the Next Generation Procurement Team” is available to download here.

---

Globality is the pioneer of agentic technology for autonomous sourcing, redefining how global enterprises buy the goods and services they depend on. Our platform’s agentic intelligence interprets purchase intent, generates RFx materials, determines suitable suppliers, and executes negotiation strategies that produce high-quality award decisions. We deliver faster cycle times, significant cost savings, and greater strategic impact from procurement. For more information visit Globality.com.

The post Agentic AI’s Transition from Promise to Proof in Enterprise Procurement appeared first on Enterprise Times.