Agentic AI
Agentic AI describes systems, often called AI agents, that plan and execute multi-step tasks toward a goal rather than returning one response to one prompt. A procurement agent might send an RFQ, chase the four suppliers who have not responded, log returned quotes, and assemble a comparison, choosing each next step itself. The defining feature is initiative inside explicit boundaries.
Examples
Quote chasing: An agent issues an RFQ to nine suppliers, sends tailored reminders on days 3 and 6 to non-responders, and answers two clarifying questions from a pre-approved FAQ. Eight of nine quotes arrive by day 8; the buyer touched none of it.
Price monitoring: An agent compares contracted extrusion prices to the aluminum index monthly. When the index drops 6% and holds for two quarters with no price movement, it opens a renegotiation task with the math attached.
Boundary in action: The same agent may release against a blanket PO within 5% of contract price on its own. A supplier's request for 9% lands in a buyer's queue within the hour instead of being quietly accepted.
Definition
The contrast with generative AI is the loop. A generative model answers when asked and stops. An agent observes state (three quotes in, six outstanding), picks an action (send reminders, rephrase the question for the supplier who misread it), executes, and checks the result, repeating until the goal is met or a rule says stop.
Agents earn their keep on high-volume, low-stakes loops nobody staffs properly: chasing quote responses, updating records after events close, monitoring contracted prices against indices, assembling status reports. Pointed at an entire sourcing event end to end, the same machinery becomes autonomous sourcing; pointed at bilateral dealmaking, autonomous negotiation.
Guardrails are not optional, because errors compound across steps: a system that is 95% reliable per step succeeds about 60% of the time across ten steps. Working deployments scope agents tightly, log every action, cap what they can commit (spend limits, approved recipients), and put human checkpoints in front of anything a supplier would read as a commitment. LightSource takes this approach: agents handle quote chasing and record updates while awards and supplier commitments stay with the buyer.
Related Terms
Autonomous negotiation
Human-in-the-loop (HITL)
Previous
*GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally, and COOL VENDORS is a registered trademark of Gartner, Inc. and/or its affiliates and are used herein with permission. All rights reserved. Gartner does not endorse any vendor, product or service depicted in its research publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s research organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.