Coupa: Powering Growth with Agentic AI & Spend Intelligence

As enterprises race to harness AI, many are finding that speed alone isn’t the secret to success. Rather, trust, explainability and measurable outcomes are the desired results.

For Coupa, the global leader in spend management, the future of AI is not just generative, it’s agentic – a new model of intelligent systems that act, reason and explain within business context. And as Peter Truman, Senior Director of Technical Architects at Coupa, explains, this approach is transforming procurement and finance into a smarter, faster, more connected function.

The pressure to deliver

The surge in AI adoption is being driven from the top. “Various organisations are driving the use of AI and wanting to see results fairly quickly,” says Peter. “We're seeing that from board level coming down into the organisation to use AI.”

But this demand for speed comes with structural and human challenges. Skilled AI professionals remain scarce, employees fear displacement and leaders face pressure to demonstrate a tangible return on investment. “It’s very difficult in many cases to actually see what that return on investment is,” Peter adds. 

“We’ve seen many organisations try to do all sorts of different things with AI and it’s really difficult to see the benefits that’s actually bringing.”

This disconnect between ambition and outcome is why Coupa is doubling down on a vertical approach to AI, says Peter, adding: “we are really focusing on the vertical approach to AI within what we’re doing within the Coupa platform.”

Agentic AI: beyond automation

The term Agentic AI has become central to Coupa’s strategy. Unlike traditional automation or generative tools, Coupa’s agentic framework – with its portfolio of Navi agents – is purpose-built for the spend management vertical.

“It really is focused on driving business outcomes across spend management processes,” says Peter. These agentic systems learn from real transactional data – billions of interactions flowing through Coupa’s US$8tn community-spend dataset – and use that intelligence to deliver recommendations, surface risks and optimise processes.

“From an agent perspective, I think it’s changing the way buyers and suppliers will interact with each other,” Peter continues. “There are rules and controls built into the Coupa platform to make sure that there’s always a human in the loop. Agents aren’t just running away doing things on their own.”

This “human-in-the-loop” design is fundamental to Coupa’s philosophy of trustable AI – a safeguard that ensures decisions remain transparent, ethical and aligned with customer expectations.

Breaking down silos and connecting the enterprise

Most organisations today suffer from fragmented systems and data silos that obscure visibility and slow decision-making. “Every organisation will have multiple applications doing multiple things, having data stuck in those and it’s quite difficult to bring that data to control the process that’s going on elsewhere,” says Peter.

Coupa’s agentic AI tackles this by connecting to other systems, retrieving relevant data – such as available budgets – and orchestrating workflows accordingly. “It’s about connecting the broad organisation, both the organisation internally and the external organisation and bringing that together,” he explains.

The platform doesn’t just automate tasks, it uses data to generate insights about supplier performance, contractual compliance and pricing anomalies. The result is a more unified, intelligent ecosystem that bridges buyers and suppliers with shared context and smarter collaboration.

At the heart of Coupa’s user experience is its portfolio of Navi agents. “Navi is becoming the user experience, taking over from the pointing, clicking and finding your way through menus,” Peter explains. By interacting with Navi, users can raise requisitions, analyse supplier data and generate reports with natural, conversational ease.

“It could be as trivial as raising a purchase requisition, or something a lot more complicated about analysing the data associated with a particular set of interactions that a Coupa customer is having with their suppliers,” he says.

Beyond routine tasks, Navi’s specialised agents – such as the network modelling agent and knowledge agent – help users simulate supply chain scenarios, extract contract metadata and reduce human error. 

“It’s bringing that higher-order work to what they’re doing rather than just automating a task that’s already there,” Peter adds. “It’s bringing in the richness of the data within the Coupa platform and helping users make decisions faster based on data they may not have even known existed.”

If AI is to gain the trust of finance and procurement leaders, it must prove measurable value. Peter is quick to caution: “AI isn’t magic. It doesn’t magically create value, it’s got to be fed with data that is relevant to what’s going on in that environment.”

But when fed the right data, agentic AI has delivered remarkable results. “It’s spotting broken approval patterns, showing supplier performance data and patterns which you may otherwise never see,” he says.

Coupa’s AI has become adept at identifying potential fraud – flagging duplicate invoices, suspicious purchase orders or attempts to skirt approval thresholds. “That for me is one of the really exciting parts where it is really enabling massive savings over potentially duplicate payments or indeed purchase orders, where people are circumnavigating the rules,” Peter notes.

Building AI businesses can trust: the ethical data moat

Trust and transparency are at the centre of Coupa’s AI development. The company has invested heavily in security and compliance to safeguard customer data. “We have invested very, very heavily in our security and compliance programme to make sure all of the data that we look after on behalf of our customers is handled and managed in a way that organisations can trust,” says Peter.

Coupa’s ethical framework extends beyond compliance – it’s about explicit data governance.

“We don’t use personal information within our AI data moat. We don’t use anything that can be traced back to a particular customer,” he explains.
The company’s AI governance programme, which includes legal oversight, ensures that development decisions are both explainable and accountable.

Coupa’s ethical data moat is one of its most powerful differentiators. Over nearly two decades, customers have granted express permission for Coupa to use anonymised transactional data – creating an unparalleled, privacy-preserving dataset.

“It’s absolutely our customer’s data, but used in a way that you can never trace anything back to a particular customer. It’s about spend management data – it’s not data that’s scraped from the internet.”

This ethical data moat enables Coupa to surface deep, trusted insights without exposing sensitive information. “For pricing insights, it will show a range. It’s not going to show that any particular customer is paying this for a particular item.

“Again, it highlights the low-risk approach of using this data.”

Coupa’s community intelligence within in spend data is the foundation for real business value.

“What we’re basically doing is using that dataset to put our customers’ data to work,” Peter says. By comparing individual data to anonymised community trends, Coupa surfaces insights into supplier risk, pricing opportunities and performance anomalies.

“If there are trends happening in the community – supplier risk, for example – it may flag that there are potential issues you might want to investigate,” he explains.

The system can also identify cost-saving opportunities: “It might show that you’re paying far too much for a particular commodity or service, while others are seeing a much lower price. That highlights a great opportunity to go through a sourcing event to get a better price.”

Driven ROI businesses can see

Peter encourages leaders to evaluate AI success through business KPIs, not abstract technical metrics. “It’s really KPIs around the business processes that people are running,” he says.

That includes sourcing efficiencies, spend savings, ESG improvements and workflow optimisation. 

Peter shares examples of Coupa customers already achieving measurable impact: “Uber looked at millions of transactions within the platform and generated tens of millions in sourcing efficiencies through contract renegotiation and supplier consolidation.”

Another success story is Microsoft: “By working with Coupa’s AI supply chain solutions, they made over a 60% cut in their transportation emissions while improving the overall efficiency of that supply chain.”

Learning to crawl, walk and run with AI

For enterprises at the start of their AI journey, Peter’s advice is pragmatic: start small, scale fast. “I wish Coupa was the magic dust that did that,” he joked. 

“It’s about focusing on specific outcomes you’re looking to achieve. Think big about the opportunity, but start with quick wins – relatively low risk, high impact areas.”

Coupa encourages a “crawl, walk, run” approach: begin with defined process improvements, then expand across the organisation. “Once you’re starting to see improvements, think about how you can scale that out fast,” he says.
The company’s “bring your own agent” capability allows customers to integrate their own AI tools with Coupa’s platform – further enhancing interoperability and innovation.

“We’re seeing some really interesting impacts of customers bringing their own agentic capabilities to Coupa and interacting between Coupa and other external systems,” Peter explains.

Peter’s final message to leaders is simple but urgent: don’t wait. “Embrace AI – study what others are achieving from AI and really embrace it,” he says.

“Look for high-impact but low-risk opportunities and just get on that journey, because people are being left behind if they’re not embracing AI.”