While procurement leaders have ambitious AI goals, most of their current systems are not ready, as outlined in The Path to AI-First Procurement: Closing the Gap between AI Ambition and Execution, released by Ivalua and Ardent Partners.
In fact, 61% of procurement professionals see AI as a productivity and scale enabler, but few know how to deploy it consistently and effectively.
“The organizations that are already scaling AI reliably are unlocking massive strategic capacity, giving them a significant competitive advantage. There is immense organizational pressure to move fast with AI in procurement, but procurement leaders must take a strategic approach, collaborating closely with IT,” says Alex Saric, CMO at Ivalua. “Chasing AI-wins on infrastructure that cannot support it delays business impact and limits scale. The playbook for success requires choosing the right unified architecture, instating human-in-the-loop controls, and only then accelerating broad deployment to make sure that AI in procurement can add real value to everyday operations.”
“Those making tangible progress with AI in procurement are not necessarily moving faster but are rather doing so with strategic intent,” says Andrew Bartolini, founder and CRO at Ardent Partners. “They are learning where intelligent systems create the greatest operational leverage, where governance matters most, and how AI can augment execution without creating new operational risks.”
Key takeaways:
· Whilst 90% of organizations are actively exploring or piloting AI, 67% are in the “overreaching“ quadrant of the report, with high AI ambitions facing underdeveloped governance and fragmented data foundations.
· Despite the challenges, 23% has successfully established “AI-first“ procurement thanks to solid data foundations, embedded governance, and integrated workflows.
· Procurement professionals are not asking AI to take over their work but rather to give them back the most impactful and strategic part of their jobs.
· Nearly half (47%) identify cognitive augmentation as their primary goal for a human-agent operating model, followed by a focus on relationships (36%).
· Data remains the most prominent obstacle, whether it’s quality, availability, or structure, cited by 59% of respondents. This is followed by 51% who mention integration with existing systems as the main challenge. Only 11% of organizations report operating a unified procurement data model across their source-to-pay process. And, 41% describe their data as integrated, yet dependent on manual reconciliations across systems, while 40% continue to operate with fragmented data and no single source of truth, and 8% have dark data trapped in unstructured formats.
· More than one in three organizations (34%) have no formal AI governance in place, with 96% of respondents viewing black box AI risk as high or moderate. Over half (51%) indicate data hallucinations as their greatest concern, followed by proprietary data being used to train public large language models (20%). And, 63% of respondents agree “human-in-the-loop“ (or mandating manual approval for critical decisions) is a non-negotiable governance requirement for procurement AI.
· Current AI use cases are concentrated in procurement domains, such as spend analytics (22%), supplier discovery and onboarding, and accounts payable automation (both at 20%). Planned deployment in the next 12 months significantly exceeds current usage, with contract lifecycle management showing the biggest gap (56% planned vs 14% in use today).
· More than 80% of organizations plan to deploy agentic AI capabilities in the next three years, but the majority of them display high ambition against low readiness as they look to deploy AI on fragmented data and underdeveloped governance.
