AI agents move into lab buying

Procurement teams are testing artificial intelligence agents to handle supplier emails, quote collection and purchase-order work that used to sit with buyers and lab operations staff. (ey.com) In procurement, an “agent” is software that does more than draft text: it can read a request, apply policy, pick a buying channel, route approvals and update systems with limited human input. EY says these systems can run parts of intake-to-invoice processing by interpreting free text or forms, selecting suppliers and following delegation rules. (ey.com) That is a fit for lab buying because research purchasing often mixes routine supplies with quote-heavy equipment orders, tax and grant rules, and back-and-forth with specialized vendors. The Food and Drug Administration’s laboratory purchasing manual says lab procurement covers supplies, materials, equipment and services that affect laboratory quality. (fda.gov) Vendors are now pitching those chores as agent work. GEP says “agentic AI” can generate purchase orders with minimal human intervention, while Zip says its platform uses purpose-built agents across intake, sourcing, compliance and spend workflows. (gep.com) (ziphq.com) The push comes as procurement leaders try to move staff off transaction processing and into planning, supplier strategy and internal advising. Kearney partner Elouise Epstein, whose work focuses on digital procurement, has argued that agentic systems could create value for both procurement teams and the broader business if companies answer governance and operating-model questions first. (kearney.com 1) (kearney.com 2) The timing is tied to a broader technology cycle in supply chains. IBM said 64 percent of chief supply chain officers surveyed said generative artificial intelligence was completely transforming their supply-chain workflows, while Deloitte said 92 percent of chief procurement officers it surveyed were planning to invest in generative artificial intelligence, but only 37 percent were piloting or deploying it in procurement at the time of its early-2024 survey. (ibm.com) (deloitte.com) Analysts are also warning that the rollout is uneven. Gartner said on July 30, 2025, that generative artificial intelligence for procurement had entered the “trough of disillusionment,” with some early adopters reporting gains in efficiency and insights while others struggled to show consistent returns on investment. (gartner.com) That gap between promise and practice is why webinars and vendor demos are centering on controls as much as automation. SupplyChainBrain promoted an April 7, 2026 webinar with Gartner analyst Magnus Bergfors on how artificial intelligence is reshaping procurement roles, processes and strategies for chief procurement officers. (supplychainbrain.com) For labs, the near-term use case is not autonomous science buying in one leap. It is software taking over the repetitive middle of the process — gathering supplier responses, drafting purchase orders, chasing confirmations and pushing updates into procurement systems — while humans keep the final say on policy, budgets and supplier choice. (ey.com) (ibm.com)