The conversation around artificial intelligence in procurement has shifted dramatically. While discussion once centred on possibilities and potential, today's procurement leaders are demanding something far more concrete: measurable return on investment. 

This evolution represents a turning point for the profession: one that requires careful navigation between technological enthusiasm and practical business outcomes.

Tricia Miller, SVP Product Marketing and Chief Evangelist at Coupa, brings a unique perspective to this challenge. With a background in procurement before moving into her current marketing role, she understands the operational realities facing procurement teams and the transformative potential of AI technologies.

Reframing the ROI question

The most critical shift in thinking about AI adoption involves reframing the question of return on investment. Rather than asking what ROI AI will deliver in isolation, Tricia believes organisations must think more holistically about their procurement objectives.

"The focus should not be on what is the ROI that AI is going to deliver,” she explains. “It’s much more ‘what is the ROI that the procurement function is going to deliver and how does AI help that?’

"These two are not mutually exclusive. AI is another tool in the belt of the procurement function that should allow them to operate more effectively or operate with a greater level of intelligence."

This perspective transforms AI from an end goal into an enabler of broader procurement outcomes. The results can be substantial. Tricia notes that, with some of the agents Coupa has recently delivered, organisations have seen reductions in manual auditing efforts to identify fraudulent buying, while efforts to create requisitions or respond to knowledge questions have decreased by between 50-70%.

She adds: "These are key outcomes that the procurement team need to deliver to show value to the organisation and they can be accomplished because they're executing with a greater level of intelligence.”

Matching AI investment to maturity levels

One of the most common pitfalls in AI adoption stems from a mismatch between technological investment and organisational maturity. 

Procurement functions exist at vastly different stages of development and successful AI implementation requires honest assessment of where an organisation currently stands.

"We as procurement leaders have to be very thoughtful about where we are on that maturity journey and then make the right bets based on that," Tricia emphasises.

For organisations at stage one of the maturity journey – characterised by tactical and operational activities – the priority should not be sophisticated category strategy creation. Instead, these organisations benefit most from AI applications in ordering, invoice processing, intake and orchestration.

As organisations progress up the maturity curve and develop a sourcing focus, AI applications shift accordingly. Strategic sourcing agents that increase efficiency or tools that streamline tactical sourcing activities become more relevant.

“Recognising where your inefficiency is, as a function, and recognising which traditional AI, (like) intelligent predictive AI, or agents, can help you address the gaps that you have, will ensure that you're getting the most value and bang for the buck sooner rather than later,” Tricia continues. 

Overcoming barriers to AI success

Even with appropriate strategic focus and maturity alignment, organisations face significant barriers to realising AI's full potential. Recent research from Coupa's AI Clarity report identifies three critical gaps that procurement leaders must address.

The first is a strategy gap rooted in a disconnect between decision-makers and actual AI users. The study found that only 5% of executives making decisions about AI deployment use AI tools daily, compared with more than 50% of technical teams.

"We have a strategy being set by potential leaders who don't fully understand the art of the possible," Tricia observes.

The second barrier is a trust gap. Half of organisations lack the AI governance frameworks necessary to set them up for success. Without clear policies, oversight mechanisms and ethical guidelines, organisations struggle to deploy AI with confidence.

Perhaps most critically, organisations face an infrastructure and skills gap. Only 34% of organisations report adequate preparedness with data structure, while a mere 21% feel confident about skills readiness.

Tricia says: "The role that a skilled, informed business and IT person is going to play in the success is critical if the organisations are going to go past those barriers.”

Finding quick wins

Building comprehensive AI capabilities takes time, but procurement leaders understandably want to demonstrate value quickly.

The key lies in identifying areas where AI can deliver tangible benefits in relatively short timeframes, matched to organisational maturity.

Tricia frames this opportunity clearly: "It is less about ‘procurement roles are going to be eliminated, procurement work will be eliminated’. Roles have a massive opportunity to be elevated within the organisation and the procurement activities that they support."

For organisations at tactical stages, smart intake and orchestration systems remove friction from processes. 

Tricia explains: "Smart intake and orchestration – ensuring that you're removing friction from the process and getting the request from the initial point of entry to where it needs to go within a procurement organisation or straight through to purchase” – will deliver value quickly.

Keeping humans in the loop

The phrase "human in the loop" has become ubiquitous in AI discussions for good reason. 

In procurement specifically, maintaining human expertise and oversight isn't just ethically important, but strategically essential.

Tricia's position on this is unequivocal: "I think human in the loop in procurement is a critical piece of the capability."

Rather than replacing procurement professionals, AI creates opportunities to elevate their roles by handling tactical work and freeing capacity for higher-value strategic activities.

"Elevating the procurement team by taking away some of the tactical work so that the procurement team is able to focus on the work that matters most is a real opportunity for organisations," Tricia contends. 

Research from Coupa's Clarity report reinforces this preference, with 65% of respondents favouring AI-assisted models with human oversight. This approach allows procurement professionals to leverage AI efficiency while applying their expertise to interpreting results, engaging with business stakeholders and driving strategic initiatives.

The transition requires thoughtful change management. Tricia emphasises the importance of “bringing procurement along on this journey, demonstrating how the team will play a role after AI drives efficiency and going one step further by fundamentally showing them how they'll support the upskilling that might be needed so that they can focus on that work that matters most”. 

Building the 2026 AI roadmap

As organisations plan their AI strategies for 2026 and beyond, Tricia offers several guiding principles grounded in practical experience.

First and foremost, AI cannot be viewed in isolation. It must be integrated into broader functional priorities, with clear connections to how it improves efficiency, effectiveness, or intelligence in pursuing organisational goals.

"AI cannot be looked at in silos,” she asserts. “It has to be, ‘what is your functional priority and how is AI going to make the accomplishment of that priority better?’" 

Starting small makes sense, particularly for organisations with immature data. 

Tricia recommends: "Start by automating elements, taking some of the agents. Start small and then communicating value. You have to pause to say, ‘is it working?’" 

Focusing on specific agents that can deliver concrete outcomes allows teams to build confidence and demonstrate value before scaling up investments.

Using AI to support decisions rather than replace them represents another key principle. An agent that summarises RFI responses, for example, can quickly assess all submissions against defined criteria, enabling procurement professionals to demonstrate how they have identified the supplier that will drive most value.

Finally, organisations must build what Tricia calls a "sandbox culture" to adapt to rapid technological innovation.

She emphasises that the era of change happening only during massive, point-in-time transformation programs is over; moving forward, change will be a constant reality.

To extract the most value from this evolution, business functions must stay closely connected to new capabilities.

This requires a specific skills mix, as Tricia explains: "We have to have our operators within the function, we have to have AI within the function, and then we have to have people partnering with IT who understand where the tech is going, and how those enhancements can be leveraged to make the organisation better."