Autonomous purchasing bots that negotiate deals at the speed of light. Programs that summarise 100-page contracts in seconds. Analytics engines that analyse and deliver new strategies on the fly as events unfold.
The future of procurement? No. That’s where we are right now. AI is transforming procurement much faster than anyone expected, and unusually, it’s happening ahead of other corporate functions. The leading multinationals we work with – and our Digital World Class® Procurement research – confirm that bringing together a function that depends on the mastery of commercial detail with a technology that excels at data analysis and synthesis is producing remarkable results at their companies:
- Autonomous tenders now channel 20-30% of spend while increasing ROI.
- Labour costs have dropped, as technology means they need 31% fewer employees, with some organisations now spending more on technology than labour in some processes.
- Procurement teams have double the ratio of strategic to operational colleagues compared to the average team.
There are concrete examples of use cases already delivering results, such as implanting autonomous sourcing at scale, using AI agents for contract analysis and supplier and should cost intelligence, and replacing traditional reporting with conversational analytics.
Adoption is lagging due to strategic and operational gaps, not technical limitations. Recent developments like Anthropic’s enterprise agent platform and plug-ins (released January and February 2026) have further upended the landscape, offering orchestration capabilities that democratise agent technology and expand automation potential, while challenging reliance on expensive SaaS tools. The growing number of affordable user licences for leading AI models means many more cost-effective and rapid solutions are possible.
Where is this heading? Pilot programs have proven that AI can achieve equal or better cost savings than human buyers. The purchasing analysts we talk to are becoming very used to letting AI review contracts or roll out assessments of suppliers in seconds. Eventually, we believe that as much as 90% of third-party spend will be handled by autonomous agents.
But that doesn’t mean the lights are going out in the procurement department. Instead, there will be many challenging projects that require sensitive and highly skilled professionals, alongside a growth agenda for all the valuable activities we never had time for. These projects will have high-value, relationship-driven objectives that are more about creating value than cutting costs. Colleagues working on these projects will use AI, but in a different way: to help them add depth and speed to their understanding and keep their deal flow on course.
When we asked procurement executives how they imagine their function will change by 2030, most of them share our view. They foresee growth in roles enabling the digital workforce (e.g. AI agents) and advanced data analytics, e.g. LLMs. They also foresee a major rebalancing away from operational roles such as managing tail-spend and accounts payables to more strategic roles like business partners, and reshaping category manager, sourcing manager, and supply manager roles to focus on strategic, high-value activities only.
The technology stack will also change. Although LLMs, AI agents, and newly orchestrated agentic workflows enable teams to look outward more than they ever have, we think procurement teams may also find themselves rebuilding walls at the same time, through moving bespoke large language models off the public cloud and back onto private clouds or on-premises to keep data secure and protect confidentiality and competitive advantage.
The jagged frontier
The future of AI-powered procurement is less clear. As Ethan Mollick, an associate professor of innovation at the University of Pennsylvania’s Wharton School and a well-known AI researcher, likes to put it, AI technology is now “a jagged frontier”, in which some elements are much more advanced than others. Several big questions remain unanswered:
- How do we capture the tacit knowledge of our experts?
- What skills will the next generation of procurement specialists need, and how should we train them?
- What exactly is it that humans are bringing to the party? Are there ways to enhance our unique capabilities?
At this point, however, the overall speed at which the field is evolving is really the most important thing to understand. For CPOs doing business in 2026, the energy and velocity of development suggest the biggest risk is not rolling out the wrong system but hesitating to join this revolution. The technology is sufficiently mature and offers superior advantages in specific areas; the biggest obstacle to success is organisational reluctance and lack of confidence to adopt rather than the systems themselves.
Fortunately, the leading companies are only a year or two ahead, so the adoption gap is not insurmountable. Organisations can make immediate progress by quickly deploying accessible tools, setting up AI agents, embracing advanced prompting techniques, and establishing AI specialists per spend category. The harder work comes from establishing AI-ready data foundations, adding funding for autonomous tools for tail-spend pilots, and orchestrating agent workflows with appropriate governance.
To achieve sustainable competitive advantage and maximise ROI from AI investments, your programme needs to identify, evaluate and design AI solutions based on your company-specific context. This means based on your process requirements, your existing automation footprint and your data source requirements. For example, use AI to act on should-cost and category intelligence in real time or stay ahead of supply resilience risks to drive value in a volatile environment.
Addressing workers’ fears of redundancy head-on is also important and best handled by emphasising the role growth opportunity and encouraging frank discussions about what is and is not working in the rollout, regardless of whether the source of friction is the team, the technology, or the vendor.
When it comes to employee education, plenty of good-quality, low-cost training opportunities are available online through Coursera, LinkedIn, and the AI model builders. The basics of AI are easy to learn, and understanding prompt construction and agents can save your team a lot of time.
It’s also important to keep up with the news and what your existing incumbent vendors are providing and planning. Although it can be difficult to separate hype from reality, significant advances are being made in AI almost every month right now. You should try to keep up with the headlines at least, because the underlying skills and capabilities are changing, and with capacities expected to double and treble over the next few years, the pace is unlikely to slow down anytime soon.
The crucial commitment CPOs need to make is to start working with the new tools immediately, rather than waiting until the benefits of using AI are crystal-clear, as that will be a point by which your team will be too far behind to catch up.