AI in Procurement: A Realistic Assessment
What I’ve learned about harnessing artificial intelligence in procurement whilst avoiding the hype trap
After spending the past few years watching AI evolve in procurement and now being more involved in panels, webinars, events, demos with suppliers and even playing around with the Claude API, I can tell you that we’re at a fascinating crossroads. The statistics are compelling: 94% of procurement executives now use generative AI at least once a week, with usage jumping 44% from 2023 to 2024. But beneath these impressive numbers lies a more complex reality that I’ve witnessed firsthand.
Through my experience implementing procurement processes and systems whilst observing the AI landscape, I’ve seen that everyone is treating AI as a silver bullet, when it’s just an additional tool. Whilst AI promises genuine transformation, the path forward requires navigating through scepticism, resource constraints, and implementation challenges that many aren’t discussing openly.
What I’ve discovered:
- Current AI capabilities deliver approximately 70% usable output requiring human refinement
- Real value creation happens once you create transparency in data processing, administrative automation, and analytical enhancement
- Success demands strategic implementation focused on business outcomes rather than technological novelty
The Collective Sigh I Keep Hearing: AI Fatigue Is Real
In every procurement conversation I have these days, mentioning “AI” increasingly triggers not excitement, but a collective sigh. Everyone talks about it, but there’s hardly a demonstration.
Our industry has weathered numerous technological storms where impressive demos failed to deliver meaningful results in real-world conditions. The shadows of past disappointments, from early e-procurement platforms to blockchain initiatives, still loom large over every AI discussion I’m part of.
I’m seeing many organisations fast-track AI adoption compared to other IT investments, yet many struggle to connect these investments to measurable business outcomes.
The 70% Reality We Live In
Current generative AI capabilities represent what I call “70% usable output”, a significant improvement from where we were just two years ago, but still requiring substantial human oversight. I’ve experienced this firsthand: AI has evolved from producing unusable results to generating information that serves as a strong foundation, but it still needs my expertise to refine and enhance.
This ~70% threshold perfectly captures where I see us today. For my procurement work, it means AI can eliminate the blank page problem and accelerate routine tasks, but it can’t replace the strategic thinking, relationship management or complex negotiation skills (yet) that I’ve developed over years of experience.
Where I’m Seeing AI Actually Work
From my observations, current AI applications (mostly ML, GenAI) in procurement cluster around 5 areas. The first three are, for most part, not really embedded yet, but used as stand alone, with a lot of copy-paste involved. Whereas with the last 2 we are seeing solutions already being in place:
Content Creation and Communication: From drafting emails that used to take half an hour to suppliers, to crafting messages for internal stakeholders, automating RFI/RFP/RFQ generation, and creating policy documentation.
Light Research: You don’t know what the requester is talking about? Feed it into your Copilot, Gemini, Claude or Chatgpt and you will get it broken down for you. You need a new supplier? With the right prompt you will find at least somewhere or someone to start with on Perplexity and the other engines.
Administrative Heavy Lifting: The most immediately valuable application I’m seeing is AI’s ability to eliminate tedious groundwork that traditionally consumes hours of analyst time. Instead of manually extracting data from poorly formatted supplier spreadsheets or cross-referencing email threads to understand contract modifications, AI handles this grunt work in minutes.
Data Analytics and Insights: Piloting or deploying Generative AI in spend dashboards, providing enhanced visibility into spending patterns and supplier performance; something I find genuinely useful for strategic decision-making.
Process Automation: Contract management, cross-checking documentation, intelligent workflows and orchestration; where systems connect, communicate and coordinate across multiple platforms and tools. These solutions combine RPA and API-based automation to streamline processes and reduce manual effort.
What this means in practice: I’m watching procurement teams use AI to parse through hundreds of supplier emails to extract key commercial terms, standardise messy spend data that would typically require weeks of manual cleanup and extract obligations from contract documents that previously required line-by-line review.
The real impact lies in removing the stupid barriers that prevent procurement managers and analysts from doing actual analysis. The time savings are immediate and measurable, often freeing up 40-60% of an analyst’s time for higher-value work.
The Challenges Procurement Is Confronting Today
Data Quality & Security Problems
Data quality has emerged as the biggest internal barrier to AI in procurement. I’m constantly dealing with concerns about poor data leading to inaccurate outcomes and misinformed decisions. SMEs in general have low data infrastructure maturity, with less than 70% of spend data stored in one place. This fragmented landscape creates significant obstacles for AI implementation, because these systems need comprehensive, clean, and consistent data to function effectively.
Equally important, data privacy and security represent big operational risks when handling sensitive supplier information, contract terms and competitive intelligence. Read the documentation (DPAs, Terms of Service, etc) and make those risks together with your Information Security & Legal team transparent.
The Skills Gap
The procurement industry faces today a critical shortage of AI-capable talent. You are in your mid 20s to 30s, no problem. Technology is less of an issue. But for everyone above that is a significant barrier to start embracing AI, play with it, test it, build your own thing with the aid of your AI engine. In general, people are too scared to use it.
This creates a double challenge for me: I need people who understand both procurement processes and AI capabilities, and that combination remains scarce in the market.
Budget Limitations
In the VUCA world we live in today, if there’s going to be money pumped into novel technology, it’s going to be in Sales and R&D. Procurement functions typically receive lower priority for transformational technology investments compared to revenue-generating departments.
When I see different new AI-native providers demonstrating how they could automate 50-80% of current procurement work, I get excited. But then I remember the funding reality: Finance and Operations VPs must compete with product development and customer-facing technologies for limited AI investment budgets.
The Opportunities I’m Excited About
Despite current challenges, AI presents very good opportunities for procurement transformation, but these require significant work from teams and providers plus executive-level commitment. Agentic AI (systems that make decisions rather than just offer information) is emerging rapidly.
These are the opportunities I see developing over the next 2-3 years, still with human oversight:
- (Semi) Autonomous tail spend sourcing: AI agents will handle the entire sourcing process for low-value, high-volume purchases with very little human intervention; automatically identifying requirements, finding suppliers, conducting negotiations and awarding contracts for routine items.
- Predictive supply disruption alerts: AI systems will analyse multiple data streams to predict potential supply chain disruptions weeks or months in advance, giving procurement teams time to activate backup suppliers before problems hit.
- Contract value leakage detection: AI will continuously monitor contract performance to identify when negotiated benefits aren’t being captured. Think of missed volume rebates, unused service credits or suppliers not meeting agreed pricing terms.
- Multi-agent workflow orchestration: Different AI agents will collaborate across the procurement process. Picture one agent identifying needs, another sources suppliers, a third handles contracts and a fourth managing performance. These agents will communicate with each other and human teams to orchestrate complex procurement workflows seamlessly.
- Real-time spend optimisation recommendations: AI will analyse spending patterns as they happen and suggest immediate actions, redirecting purchases to preferred suppliers, flagging maverick buying, or recommending contract consolidations.
- Intelligent Negotiation: Emerging pilot scenarios involve negotiation systems being deployed for tail spend and contract renewals. These systems can be configured with specific goals (e.g., 10% price reduction) and programmed with conversational strategies that use pattern matching to mimic emotional responses. While I cannot imagine human-to-bot negotiation becoming standard for strategic spend, we might see more experimentation with bot-vs-bot scenarios as early proof-of-concepts.
How I’m Approaching Implementation
After watching teams struggle with the same fundamental issues around AI, I’ve developed a framework that actually works. The reality is that most procurement AI failures aren’t technology failures, they’re strategy and foundation failures. With the right technical expertise and a disciplined approach, we can solve the core problems that derail most AI initiatives.
My North Star Strategy
The difference between successful AI transformation and expensive experiments lies in one thing: starting with business problems, not technology solutions. I anchor every AI initiative in measurable business objectives that procurement leaders actually care about.
Most organisations approach this backwards. They select AI tools first, then try to find problems to solve. I reverse this entirely: define the business challenge, measure the current state, then architect the right AI solution.
Want to cut through the AI hype? Download our 30-Day Procurement AI Reality Check.
Test AI on your actual work, not vendor demos. In 30 days, you’ll know exactly what AI can and can’t do for your procurement with real data. Includes tested prompts, cost calculators, and warning signs of AI snake oil.
How I Measure Success and ROI
Here’s what separates my approach: I track the same KPIs that matter to any procurement executive, because AI success should be measured by business impact, not technical metrics:
- Direct Cost Savings: Measurable reductions in procurement spend with clear attribution to AI interventions
- Process Efficiency: Cycle time improvements and administrative cost reductions that translate to FTE savings
- Quality Improvements: Reduced errors, better supplier compliance, and enhanced risk management with quantified impact
- Strategic Value: Innovation enablement, sustainability improvements, and competitive advantages that support broader business objectives
The breakthrough insight: these ROI metrics aren’t any different from any other procurement project. AI isn’t special, it’s just another tool that either delivers business value or it doesn’t.
My Path Forward
The journey toward AI-enhanced procurement requires balancing technological opportunity with practical implementation realities. I’ve learned to develop strategies that acknowledge both the promise and limitations of current AI capabilities. The 70% usability threshold represents significant progress, but also highlights the continued need for human expertise and strategic thinking.
The organisations that will succeed are those that approach AI implementation by studying the landscape carefully, understanding their organisation’s capabilities, and maintaining respect for the complexity they’re attempting to manage. AI is a powerful tool for augmenting human capabilities rather than replacing them entirely.
As the procurement function evolves, leaders who can distinguish between genuine opportunity and marketing hyperbole will position their organisations for sustainable competitive advantage. The transformation through AI is real, but so too is the need for skilled leadership to navigate these waters successfully.
For C-Suite Leaders: Anchor AI initiatives in business outcomes rather than technological capabilities. Invest in data infrastructure as a prerequisite for success.
For Procurement Directors: Focus initial efforts on administrative automation and data processing applications. Establish pilot programmes with clear success metrics.
For Operations Managers: Identify high-volume, low-complexity processes suitable for AI automation. Establish data quality standards and user training programmes.
The transformation is real, but success demands more than technological enthusiasm. It requires the kind of strategic thinking and realistic expectations that come from hands-on experience with both the promise and the limitations of AI in procurement.
This analysis draws upon my experience in procurement transformation and AI, combined with industry research and case studies, to provide realistic guidance for AI implementation. All statistics and findings are cited from verified, recent sources representing leading research in procurement.
References and Further Reading
- The Hackett Group (2025). Embracing the Future: How Generative AI Is Revolutionizing Procurement in 2025. Key finding: 64% of procurement leaders expect fundamental change within five years, with 49% piloting AI initiatives achieving up to 25% productivity improvements. Available at: https://www.thehackettgroup.com/insights/embracing-the-future-how-generative-ai-is-revolutionizing-procurement-in-2025/
- Art of Procurement (2025). State of AI in Procurement in 2025. Key finding: Weekly use of generative AI within procurement increased 44 percentage points from 2023 to 2024, with 94% of executives now using AI weekly. Available at: https://artofprocurement.com/blog/state-of-ai-in-procurement
- Deepstream (2025). Unlocking AI in Procurement: Six Key Takeaways from Industry Experts. Key finding: Provides practical insights on avoiding AI hype fatigue and implementing realistic use cases based on expert panel discussions and real-world procurement experiences. Available at: https://www.deep.stream/blog/unlocking-ai-procurement-six-key-takeaways
- KPMG US (2024). How Generative AI Will Transform Procurement. Key finding: KPMG simulations show AI can automate 50-80% of current procurement work, with significant potential for strategic value creation. Available at: https://kpmg.com/us/en/articles/2024/how-gen-ai-will-transform-procurement-as-we-know-it.html
- McKinsey & Company (2024). Revolutionizing Procurement: Leveraging Data and AI for Strategic Advantage. Key finding: Sanofi achieved 10% spend reduction and 281% increase in negotiation savings through AI implementation. Available at: https://www.mckinsey.com/capabilities/operations/our-insights/revolutionizing-procurement-leveraging-data-and-ai-for-strategic-advantage
- Deloitte US (2024). Generative AI in Procurement. Key finding: 92% of CPOs are assessing GenAI capabilities, with investment plans doubling from $1M+ annually. Available at: https://www2.deloitte.com/us/en/blog/business-operations-room-blog/2024/generative-ai-in-procurement.html
- Procurement Magazine (2024). What Impact Will AI Have on Procurement in 2025. Key finding: AI could improve compliance by up to 100% by 2025, with 56% of executives prioritizing source-to-contract applications. Available at: https://procurementmag.com/technology-and-ai/what-impact-will-ai-have-on-procurement-in-2025