Key Takeaways
- Gartner predicts that by 2028, 90% of B2B transactions will be intermediated by AI agents, channeling over $15 trillion in procurement spending through agent exchanges
- SAP's Joule bid analysis agent and Pactum's AI negotiation platform demonstrate that procurement automation has already reached the practical deployment stage
- Agentic AI adoption among B2B suppliers remains at just 24%, with ERP modernization delays forming the primary bottleneck
The Agentic Commerce Shock Transforming B2B Procurement
In November 2025, Gartner released a striking forecast: by 2028, 90% of B2B transactions will be intermediated by AI agents, channeling over $15 trillion in procurement spending through agent exchanges. The future where traditional human-to-human negotiations and phone-and-email-based ordering give way to machine-to-machine transactions is approaching faster than anyone anticipated.
This transformation goes far beyond operational efficiency. Viewed through the lens of agentic commerce, B2B procurement stands as one of the domains where AI agents deliver the greatest impact. B2B transactions involve repetitive negotiations, complex contract comparisons, and massive volumes of document processing. These are precisely the tasks where AI agents can outperform humans.
So what exactly is happening? Let us examine how agents function at each stage of the procurement process and what results they are achieving.
A $15 Trillion Market Restructuring: Why B2B Procurement Leads the Charge
While B2C agentic commerce captures headlines with chatbot-powered product recommendations and one-click purchases, B2B dominates in sheer business impact. Global B2B transactions total tens of trillions of dollars, with the vast majority still dependent on manual processes.
McKinsey's analysis indicates that procurement departments deploying agentic AI can achieve 25 to 40 percent efficiency gains. Individual category agents deliver 15 to 30 percent efficiency improvements through the automation of non-value-added activities. These figures dwarf the scale of B2C personalization effects.
Why are such dramatic improvements possible? Consider the typical flow of B2B procurement: supplier discovery, RFP creation and distribution, quote comparison and analysis, contract term negotiation, purchase order generation, and delivery tracking. Human buyers spend hours to days on each of these steps. AI agents compress the entire workflow to minutes or hours.
| Capability | Traditional Procurement | AI-Assisted Procurement | Agentic Procurement |
|---|---|---|---|
| Supplier Selection | Manual RFP creation and evaluation | AI scores and ranks candidates | Agents autonomously discover and evaluate |
| Quote Comparison | Manual spreadsheet analysis | AI calculates total cost automatically | Multi-agent instant negotiation |
| Purchase Orders | Manual ordering via approval flows | AI recommends, humans approve | Auto-order within thresholds, humans approve exceptions only |
| Delivery Tracking | Individual follow-ups via email/phone | Centralized dashboard monitoring | Predictive risk detection with auto-sourcing alternatives |
| Time Required | Days to weeks | Hours to one day | Minutes to hours |
As this table illustrates, agentic procurement is moving beyond the stage where "AI recommends and humans decide" toward a phase where agents autonomously complete transactions within defined parameters.
Automating Supplier Negotiation: Pactum and SAP Ariba in Action
The highest-impact area within the procurement process is the automation of supplier negotiations. An era has arrived in which negotiations that once relied on the experience and intuition of seasoned buyers are now executed through AI agent-to-agent dialogue.
Pactum's case provides the clearest illustration of this shift. The company's AI agents execute negotiations with hundreds of suppliers simultaneously, guided by policies and strategies set by the procurement team. By processing in parallel what human buyers previously handled one at a time, agents dramatically improve both coverage and speed. Forrester predicts that procurement teams will deploy agents capable of "scaling negotiation across hundreds of suppliers simultaneously," transforming static price lists into dynamic negotiation interfaces.
SAP, with its enterprise procurement foundation, has taken a different approach. The next-generation SAP Ariba announced in March 2026 deeply integrates the AI assistant Joule into procurement workflows. The standout capability is its bid analysis agent. In complex bid scenarios, it automatically evaluates not just unit prices but shipping costs, payment terms, and total cost of ownership, presenting the optimal options. Analysis that once took days in spreadsheets is now completed instantly by the agent.
These two examples demonstrate that the agent-to-agent era has already reached practical deployment in B2B procurement. A buyer-side agent queries a seller-side agent for pricing, inventory, and delivery terms, and both systems automatically reach agreements within defined rules. Human approval occurs only at the final stage, and many routine transactions are fully automated.
As PYMNTS's analysis points out, B2B marketplaces are no longer mere intermediaries but are transforming into infrastructure that enables agent-to-agent transactions. Platforms integrating procurement, logistics, and payments are beginning to function as the foundation for fully automated trade.
The Collapse of Information Asymmetry and the Democratization of Pricing
Another structural shift triggered by agentic procurement is the elimination of information asymmetry.
A PYMNTS report describes how "AI killed information asymmetry in B2B procurement." Traditionally, suppliers held an information advantage regarding their own cost structures and margins. For buyers to understand broader market pricing trends, they needed to attend industry events or make individual inquiries to multiple suppliers.
AI agents are upending this structure from its foundation. Market scans, vendor comparisons, pricing benchmarks, and even negotiation scenario simulations can now be executed independently by procurement teams. This opens the possibility for small and medium enterprises to wield negotiating power comparable to that of large corporations.
This shift also forces a strategic pivot on the supplier side. Profit models dependent on pricing opacity are no longer viable. Instead, delivery reliability, quality consistency, and post-sale support quality are gaining weight in AI agents' decision-making frameworks as differentiation factors beyond price.
The ERP Modernization Barrier
Based solely on the developments described above, one might assume that the agentic transformation of B2B procurement is advancing rapidly. The reality is more nuanced.
Deloitte Digital's research revealed that agentic AI adoption among B2B suppliers stands at just 24%. This compares to 38% on the buyer side, representing a significant gap. Why does supplier-side adoption lag? The answer lies in ERP modernization.
87% of surveyed suppliers reported that they are currently upgrading or planning to upgrade their ERP systems. ERP modernization is a multi-year, large-scale undertaking. Both IT talent and budgets are absorbed first by this foundational infrastructure work, leaving little capacity for AI investment. Gartner has also warned that over 40% of agentic AI projects will fail by 2027 because legacy systems cannot support modern AI execution demands.
At the same time, clear data shows that digitally mature companies exceed annual sales targets by 110%. In other words, an irrecoverable competitive gap is widening between companies that position ERP modernization as a prerequisite for AI and steadily advance, and those that defer it.
Building the Data Foundation for Agent Readiness
Beyond ERP modernization lies the construction of a data foundation for agent readiness. For AI agents to autonomously handle procurement, clean, structured, real-time accessible data is essential.
A Digital Commerce 360 survey found that 67.2% of respondents rated order-network connectivity as "very important" or "extremely important" for enabling AI-driven commerce. This is because AI systems dynamically route orders across suppliers based on inventory availability, pricing, and delivery conditions.
What specifically is required? First, structured product catalog data. Part numbers, specifications, certification information, and compatibility data must be organized in machine-readable formats. Second, real-time inventory and pricing APIs. Agents retrieve information through live data feeds, not static catalogs. Third, connectivity with integrated platforms such as Salesforce's Agentforce.
commercetools has described the evolution of agentic commerce in B2B as moving "from efficiency to autonomy." Without a data foundation, even efficiency remains out of reach. Autonomy is the destination that lies beyond.
Looking Ahead: The Future of B2B Procurement
Measured against McKinsey's automation curve, B2B procurement currently sits at the transition from partial automation to conditional autonomy. Whether Gartner's prediction of 90% AI agent intermediation by 2028 materializes depends on the foundational work carried out over the next two years.
The priorities for procurement leaders are clear. First, align ERP modernization with the AI roadmap. Second, begin structuring supplier data for machine readiness. Third, start with small-scale pilots, such as tail-spend automation, and build a track record of success.
The agentic transformation of B2B procurement is not a question of "if" but "when it reaches your organization." First movers are achieving cheaper, faster, and smarter procurement, and their advantage compounds over time.
