As enterprises enter the final stretch of 2025, procurement finds itself standing at a structural inflection point—one that Forrester’s Senior Analyst Jeffrey Rajamani captured with precision in his PLaN 2025 session “CPO Agenda 2026: Agentic AI Matters, Why You Shouldn’t Wait!”. With 2026 just weeks away, the message he delivered was unambiguous: the era of exploratory AI is over. Agentic AI—the ability of software to plan, act, adapt, and execute autonomously—is now the defining capability separating procurement leaders from laggards.
The Adoption Curve Has Shifted From Curiosity to Commitment
Rajamani began by grounding the conversation in fresh data from Forrester’s Q1 2025 Agentic AI Survey, showing a dramatic acceleration in planned deployments. Nearly half of surveyed AI and transformation leaders expect closed-loop agentic systems to be implemented across 25% of enterprise departments within the next 12 months. Another 29% expect widespread adoption by 2027.
What stands out isn’t just the percentage—but the pace. Enterprises are collapsing what once would have been multi‑year AI roadmaps into an 18–24 month horizon. This is driven by three converging market realities:
- Expectations from the business have changed. Leadership teams no longer view AI as a “future enabler.” They expect present‑day outcomes—faster cycle times, better compliance, higher throughput, improved forecasting accuracy, and reduced operational drag.
- Traditional procurement systems have reached saturation. The combination of RPA, static workflows, dashboards, and rule-based engines cannot keep pace with modern enterprise complexity.
- Agentic AI has matured faster than anticipated. As Rajamani emphasized, we have entered the stage where AI doesn’t merely respond—it acts. It takes decisions, adapts to ambiguous environments, and executes multi-step workflows autonomously.
What Exactly Makes Agentic AI Transformative?
Rajamani outlined Forrester’s formal definition of agentic AI: “systems of foundation models, rules, architectures, and tools that enable software to flexibly plan and adapt to resolve goals by taking action in their environment, learning, with increasing levels of autonomy.”
He distilled its three core characteristics:
- Goal orientation: Instead of being instructed step-by-step, agents understand objectives (e.g., “source this requirement,” “classify this spend,” “evaluate these bids”).
- Autonomy: Agents take novel actions, follow reasoning paths, and adjust strategies without predefined workflows.
- Tool usage: Agents leverage APIs, RPA bots, data services, supplier networks, and visualization tools, whatever is required to execute the goal.
This distinction matters deeply for procurement. Compared to RPA and GenAI, agentic AI is the only paradigm capable of combining reasoning, action, context management, real‑time adaptation, and multi-system orchestration.
Agentic AI Is Procurement’s First True Intake-to-Outcomes Technology
Rajamani walked through a compelling infographic mapping agentic AI against the full Source-to-Pay lifecycle—intake, sourcing, contracting, SRM, risk, invoicing, and analytics.
This reflects a major shift: procurement’s pain points are not isolated tasks but fragmented, cross-functional journeys. Intake requests sit incomplete. Contracts take weeks to interpret. Supplier messages pile up in inboxes. Tail-spend events stall due to bandwidth shortages. Each interruption compounds delays.
Agentic AI moves procurement away from task automation and toward outcome automation.
During his PLaN session, Rajamani reinforced a point that resonated across the global audience: agentic AI behaves like an active procurement teammate, not a passive automation engine. It interprets business intent, manages context, engages suppliers, follows policy, and keeps processes moving without hand-holding. This is the closest the function has ever come to scalable, digital co-workers.
The Use Case Universe Is Broader Than Most Realize
Forrester has mapped 15+ high-impact procurement use cases—from supplier scouting and risk analysis to negotiation, spend classification, CLM automation, and demand forecasting. What makes these use cases uniquely suitable for agentic AI is their blend of structured tasks, frequent ambiguity, and heavy need for interpretation.
A few that stood out:
- Autonomous negotiations: Agents can parse supplier emails, compare quotes, generate counteroffers, and guide award decisions.
- Supplier evaluations and SRM: Agents can monitor performance data, highlight anomalies, and surface risk insights proactively.
- Strategic sourcing orchestration: Agents can manage multi-round sourcing events end-to-end.
- Intake triage: Agents interpret intent, evaluate policy fit, and auto‑route requests to the right path.
Many procurement leaders underestimate how much time is lost in manual orchestration and “process babysitting.” Rajamani’s view: agentic AI is built for exactly these inefficiencies.
ROI: Faster, Cheaper, More Compliant, More Predictable
Jerry’s analysis shows that the financial case is unmistakable. The ROI categories highlighted—cycle-time reduction, productivity lift, spend visibility, improved compliance, and better quality of work—reflect both hard and soft benefits.
Notably, Forrester positions agentic AI in the medium-term benefit horizon (2–5 years), meaning enterprises that start now will hit peak value between 2027–2030.
But Rajamani offered two deeper insights:
- The biggest ROI is time. As he put it, “the real benefits are those hidden benefits—saving time.”
- Agentic AI compounds value. Agents learn and refine themselves as they execute tasks, improving outputs over time.
This is distinct from static no-code workflow tools, which degrade if not manually updated.
Pitfalls Procurement Must Avoid
Rajamani also provided a grounded set of cautionary notes, beginning with a visual metaphor: “AI is (still) all about the data.” Without high-quality data, organizations end up building a “house of cards.”
Key pitfalls discussed:
- Choosing use cases that don’t pass feasibility or ROI benchmarks.
- Expecting immaculate data before starting, slows adoption unnecessarily.
- Underestimating the cultural shift required when humans begin managing agents.
- Lacking trust mechanisms, including explainability, governance, guardrails, observability, and bias control.
These are essential safeguards, especially as agent autonomy increases.
What Procurement Leaders Should Do in the Next 90 Days
Rajamani closed with a clear playbook:
- Start small but demonstrate ROI quickly.
- Adopt a “fail fast, fix faster” mindset.
- Build trust through transparency and change management.
- Encourage teams to become managers of agents.
- Choose high-impact use cases first.
This guidance pairs perfectly with Zycus’ Merlin Agentic Platform and its Intake-to-Outcome architecture, which allows enterprises to deploy outcome-oriented flows across intake, sourcing, negotiation, contracting, and invoicing.
The Window Is Narrow—and Procurement Cannot Afford to Wait
If there was one takeaway from Jeffrey Rajamani’s PLaN session, it is this: 2026 will be remembered as the year enterprises operationalized agentic AI, not explored it.
Organizations that delay risk:
- falling behind business expectations,
- missing the enterprise-wide AI alignment,
- letting shadow AI tools emerge,
- being bypassed by business units seeking faster outcomes.
Agentic AI is no longer a conceptual trend. It is procurement’s new operating model.
The future is not approaching—it has already arrived. And the procurement teams that act decisively in early 2026 will define the next decade of performance, efficiency, and enterprise impact.
Related Reads:
- AI Agents in Procurement: A Comprehensive Guide
- Guide to Procurement Agents: Roles, Skills & How AI is Changing the Game
- Whitepaper: Beyond Integration – 8 Reasons to Choose an AI-Agent Orchestrated S2P Suite
- Solution: Zycus Merlin Agentic AI Platform
- Autonomous AI Agents in Action: The Future of Procurement
- On-demand Webinar: How AI Agents Supercharge Lean Procurement Teams
