The Difference Between Using AI And Winning With It

Salvatore Lombardo is Chief Product and Technology Officer at Coupa.

gettyWith the advances in LLMs and agentic tools, it can be tempting to think that “smart” AI is all that is needed. But without real systems of record sitting underneath them, AI tools are essentially very smart strangers, lacking grounding in the day‑to‑day reality of a business.

Here’s the reality on the ground: Every business is now operating in the AI economy. AI is changing everything: how work gets done, how decisions are made, how customers are served, and how new products are built. I’ve seen this up close across our network, which has processed more than $10 trillion in enterprise spend, as well as inside our organization. And the pace of change is only accelerating.

For the many organizational leaders who have spent the last couple of years debating the merits and urgency of implementing new, ever-more-capable AI solutions, a more practical (but no less pressing) set of questions has begun to emerge: “What are we currently able to do—that is, what are our actual, meaningful capabilities?” And “What things do we want to be able to do, but currently can’t?” And most importantly, “How should we be using these AI tools we have to capture value and sustain growth?”

Many organizations have already been leveraging AI tools at large scale—within multiple departments, or even among partner agencies—but only realizing discrete tactical gains: say, improved procurement platforms, more dynamic customer management systems or chatbots that engage more meaningfully with clients. These tools all solve important problems, but they do not, even when taken together, represent a strategic shift.

Everyone is using AI. For most, this simply means layering AI tools atop existing processes. But winning with AI isn’t about adopting AI for its own sake. It’s about building the capabilities, operating models and culture needed to turn this rapidly evolving technology into a real business advantage. The gap between using and winning is one that organizational leaders must close. But how?

Going from “using” to “winning” won’t be accomplished by declaring an enhanced AI vision. It requires building the right foundations to scale AI across the business and create lasting value:

Only when these foundations are in place can leaders move beyond applying AI to existing tasks and workflows and start asking bigger questions, such as what their functions would look like if they were rebuilt for an AI-driven world. They can make the meaningful shift where AI becomes the lens through which they reimagine work, make investments and decide which capabilities will actually matter.

Once leaders have reimagined how work gets done, the capabilities they need become much clearer. Those priorities should also shape how they evaluate AI partners: not by the number of AI features they offer, but by whether they can help deliver sustainable business outcomes.

To truly win in the AI economy, companies need to consider two important questions: With whom should they partner, and how flexible does their platform need to be? They require partners who can not only help carry the workload today but also support a pivot in the event that tomorrow’s AI stack suddenly looks different.

Within this workflow redesign, the mindset shift I push hard on is to stop thinking of software as a “tool” to use. It should instead be viewed as a system of intelligence. We are entering the era of software as a “partner that knows.”

But what leaders actually need is software that knows them. That means a platform that holds the core data for a function—procurement, finance, supply chain—and uses AI in conjunction with a living, evolving record. In this framework, a software vendor isn’t just selling features but a system capable of learning alongside the organization it serves. The software vendor also sees patterns across many organizations of similar size and shape; they understand where friction can occur, which decisions are high‑stakes and how regulations and risk profiles are shifting.

Choosing the right software partnership matters, especially as underlying business models shift. SaaS was built on paying for seats, one subscription per human user, but agentic-as-a-service flips that logic. With agents organizations are increasingly paying for outcomes rather than access. This shift raises the bar on the partner selection. The right technology provider isn’t just experimenting with the latest AI models; they’re building for security, compliance and the data foundation needed to create lasting business value, with incentives to stay aligned to the results you’re trying to achieve.

Without that alignment, organizations risk creating a patchwork of disconnected AI point solutions that introduce integration complexity, technical debt and inconsistent experiences. In practice, a good software partner is what separates a collection of clever AI features from a trusted, integrated system that can support an organization’s operations and goals.

For software to truly function like a partner that knows, it needs access to good information, which (in AI terms) means good data with context. In tech, we often hear some version of the adage: “Garbage in; garbage out.” The cliché happens to be true.

When models work with bad data, or without context, the insights they deliver are flawed and fundamentally untrustworthy. Even with correct but siloed data, models are unable to gain a clear picture of the situation on the ground.

Most organizations already have years of transactional data at their disposal, like purchase orders, invoices, contract redlines and supplier records, but they may be unsure how to use them to deliver value. Putting company data in context alongside aggregated, anonymized data from thousands of similar organizations delivers insights that go far beyond any one organization’s own experience—within an industry category, or across an entire region.

This is the “collective brain” effect. A single organization’s data is valuable, but it’s also limited and idiosyncratic. When combined (responsibly and securely, of course) with patterns from other organizations, it enables models to deliver a far better set of insights. In the context of transactional data, it can suggest better defaults, surface suppliers that perform more reliably in a given region or flag risk patterns that have tripped up similar organizations. It’s guidance grounded in what is actually happening in real time, not just one company’s history. This type of collective brain is what the right software partner can facilitate.

Clean data is only part of the solution; using data correctly is equally important. The underlying transactions have to be organized and labeled so the system can tell what’s authoritative and what’s just a reference point. Once the groundwork exists, every new interaction becomes another data point the system can learn from, making the whole collective intelligence pool a little smarter ... not just for one organization but for everyone drawing on it.

Winning in the AI economy requires far more than adding AI capabilities. It requires building an AI ecosystem that can evolve with the business.

A smart approach is to work with a vendor whose platform spans the full workflow while integrating with the trusted third-party systems already in place. And just as importantly, the platform should be able to coexist with any AI agents built outside of it, enabling them to transact securely rather than forcing organizations into a closed ecosystem.

Flexibility is what turns that partnership into something future‑proof. The right platform doesn’t lock you into only one vendor’s agents. It should let you tune and compose your own, adjusting how they behave, deciding when they bring humans back into the loop and having one agent talk to another. AI-driven processes must be transparent, explainable and capable of being unwound if needed. Flexibility translates to control. That is how your partnership becomes sustainable long-term, in an economy that will likely become ever more agent-led.

In a world where models, regulations and business needs will all keep shifting, the competitive advantage comes from this combination: a trusted software partner that hosts your system of record and a flexible AI platform that can evolve with you instead of holding you back. Take these simple questions into your next leadership meeting: Which of your functions could be rebuilt around an agent tomorrow? What’s actually stopping you from doing it?

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?