AI procurement pioneer and Globality Chief Technology Officer, Keith McFarlane, summons his crystal ball. But did he do that by typing a prompt or asking a question?
2025 was a landmark year for enterprises evaluating AI’s potential in the huge business arena of corporate procurement. As procurement can account for as much as 75% of a Fortune 500 company’s total spend, anything that could potentially optimise that process deserves the Chief Procurement Officer’s serious consideration.
The experts seem to agree. McKinsey is closing the year predicting that, when done right, AI agents can make the function “more efficient, more agile, and increasingly strategic.” The firm says that could result in the procurement function being 25 to 40% more efficient, while repurposing team activity from routine tasks to strategic decision making. Hackett Group reported that while adoption is still in its early stages, procurement leaders expect it to provide “breakthrough” levels of value, with some organisations already achieving productivity improvements of 25% or more.
On the front lines of procurement execution, major brands like Amazon Business are rolling out AI tools designed to streamline purchasing workflows, improve spend analytics, and monitor supply chains. In a recent webinar we ran, our customer Rhonda Spraker Griscti, Executive Director of Agile Sourcing at Bristol Myers Squibb, noted that thanks to AI, their RFP timeline had dropped from six-nine months to 27 days and had eliminated five months of cycle time. They are now processing ten times more RFPs than before, an outcome, she said, almost unheard of in software rollouts.
Success unquestionably does depend on how you implement, integrate, and use AI, not just the tech itself. Some organisations encounter integration challenges, uneven ROI, and data quality issues; effective implementation requires solid data foundations and strategic alignment with enterprise systems.
So, where does this leave the market going into 2026? AI in corporate procurement will be a winning use case, helping to clear massive backlogs, cutting operating expenses, delivering ROI and even improving the function’s reputation and morale. It’s a bold statement, so let me break down some of the specific trends I expect to see driving this next year.
Procurement will trend towards full automation
Procurement teams will increasingly use AI agents to do things like build and structure custom sourcing events, communicate with suppliers, and negotiate contracts. In the next 12 months, I predict that enterprise use of agents will resemble the speed of ChatGPT adoption thus far, moving ever closer to full automation of certain processes. Developments like the new MCP standard for AI to connect to multiple back-end data sources will be a big accelerator.
I also expect to see teams in 2026 using AI agents to negotiate in much more personalized styles. For example, I can see procurement professionals briefing an autonomous agent to adopt a more collaborative or more competitive tone, depending on the context. As autonomous negotiation becomes more mainstream, I expect this type of capability to mature significantly.
In line with McKinsey’s predictions, I believe automation will continue to help procurement teams move beyond repetitive tasks to focus on strategy and value creation. As a result, agents will have an almost immediate bottom-line impact: after all, many organizations face large backlogs of sourcing events that never receive proper analysis or negotiation. Delegating routine work to AI can help reduce that backlog, improve coverage, and give the CPO enviable levels of control over spend.
Governance and decision transparency will become key
Two aspects of transparency will gain prominence in 2026, beginning with governance transparency. Every company using AI will need to disclose how it manages AI, including what principles guide its use, how data is handled, and what safeguards are in place (see the EU AI Act). Larger enterprises are already demanding this from suppliers, sending detailed AI questionnaires and excluding those with weak or incomplete answers. To stay competitive, brands will need to clearly articulate their AI governance policies and back them up with evidence in the forms of audit trails, documentation, and verifiable architectural practices that ensure customer safety. The second aspect is decision transparency, making it clear where AI is involved in decision-making and explaining why a particular outcome occurred.
Building trustworthy AI? Show the receipts
It’s early days in AI, but I anticipate more governance and compliance standards to start to emerge next year. From safety to privacy, companies must prove they operate at the highest levels of benchmarked competence to build consumer trust in their AI products and services.
Adoption is still in its early stages, but over 2026 and beyond, much like ISO 27001 did for information security, I expect ISO 42001 certification to become a common badge of credibility for AI software. Full disclosure: my company is currently evaluating ISO 42001, and we see real value in the framework it provides. In response to the standard, we’ve already established an AI Governance Committee, and our InfoSec team is well-versed in its requirements.
The move from generic to vertical industry models
Growing pressure for AI investments to deliver ROI is likely to create a sharper distinction between all-purpose LLMs and vertical industry models. As LLM providers attempt to better monetise their offerings, I expect the pace of innovation in the former will slow, and we’ll see the incremental updates you’d expect from any evolving software. The LLMs you’ll see in 2026 are likely to hallucinate less, handle function calls better, and offer larger context windows for less reliance on retrieval-augmented generation (RAG), for example.
Incremental model improvements are not the limiting factor, because most users are still learning how to make effective use of the AI capabilities already available. Through 2026 and likely well into 2027, most enterprise AI progress will come from improving how we integrate and apply existing models. What might that look like? I expect enterprise AI teams to start complementing the general-purpose LLMs with smaller, specialised models that use fewer resources and are better aligned with real-world use cases. These domain-tuned models can build on the strengths of the larger ones, performing better in context because they’re trained on data grounded in actual business realities rather than broad, ungoverned internet text.
Voice will get louder
Finally, not only in procurement but across many industries and use cases, I think we’ll see a much richer set of interfaces to work with our business AIs. That could include a real shift toward using voice, and voice combined with video, for example. Generations Alpha and Gen Z mainly use voice to converse with generative AI systems already.
When you’re in a true verbal conversational flow with an AI, it changes how you perceive the interaction and how your brain responds to it. Just as brainstorming in-person with a whiteboard usually sparks more ideas than trading emails or chat messages, speaking with an AI can be more creative and productive.
That could result in much better, more natural working practices. Imagine being able to ask one of your internal experts to literally verbalise what they want to go into a sourcing event: “Using the previous document, change to a more assertive, urgent tone. Include a reference to what we agreed in our last meeting as part of the introduction.”
If all this seems too speculative, just remember only three years ago none of us were talking about ChatGPT or LLMs! So, while some areas are still too nascent to predict (AGI, I’m looking at you) the speed of innovation and value-add from AI in procurement might just exceed my outlook.
