Procurement teams have long been promised automation, yet many still find themselves reacting to supplier delays, invoice exceptions and changing delivery dates only after the damage is done. The central limitation is not a lack of workflows or dashboards, but the fact that most ERP systems record events rather than interpret them in time to prevent disruption.
That is the case being made for agentic ERP source-to-pay automation: AI agents embedded in ERP environments that can ...
Continue Reading This Article
Enjoy this article as well as all of our content, including reports, news, tips and more.
By registering or signing into your SRM Today account, you agree to SRM Today's Terms of Use and consent to the processing of your personal information as described in our Privacy Policy.
watch for supplier signals, assess their likely impact and take coordinated action across procurement processes, while still leaving organisations in control of the final decision-making.
The shift is attracting serious attention. In April 2026, Gartner said supply chain management software with agentic AI capabilities is set to rise from less than $2 billion in 2025 to $53 billion by 2030. The research firm also expects 60% of enterprises using SCM software to have adopted agentic AI features by then, up from 5% in 2025.
Traditional source-to-pay automation was built for a more predictable world, where a fixed rule could trigger a fixed response. That model works when supplier performance is stable and demand patterns are easy to forecast. It is far less effective when purchase order changes arrive by email, contract terms shift and operational decisions depend on data scattered across systems.
McKinsey has previously said procurement functions use less than a fifth of the data available to them in decision-making, with the rest trapped in inboxes, supplier portals, PDFs and disconnected applications. That gap between available information and usable intelligence is what agentic systems aim to close.
Inside Microsoft Dynamics 365 Supply Chain Management, the concept is being translated into specific tools. Microsoft’s Procurement Agent, now in public preview, is designed to monitor supplier communications, identify purchase order changes and estimate the likely effect on stock levels, production schedules and customer commitments. Farmlands Cooperative in New Zealand, which has consolidated seven ERP systems onto Dynamics 365, has said the tool now handles half of its purchase order email traffic and is expected to save around 20 hours a week.
Microsoft’s broader approach also includes Dynamics 365 ERP Model Context Protocol servers and Microsoft 365 Copilot, which are intended to let agents access live ERP data securely without forcing users to move through traditional screens. In practical terms, that means a procurement manager reviewing a bid can bring in vendor history, supplier correspondence and transactional data in one place before awarding work.
The financial implications extend beyond procurement alone. Microsoft’s Finance Agent, built on Business Performance Analytics in Dynamics 365, is designed to help leaders examine how sourcing choices affect margins, cash flow and working capital using natural language queries.
Even so, the technology is not presented as a complete answer. Microsoft and its partners are positioning agent-ready ERP as a flexible foundation rather than a closed system. Partner-built tools such as Sonata’s Vendor Onboarding Agent, MCA Connect’s Smart Sourcing Agent and KPMG’s Supplier Insight Agent tackle specific pain points, while Copilot Studio allows organisations to build their own agents for specialist approval chains, compliance requirements or unusual supplier networks.
The common thread is governance. Industry analysts have warned that the rapid spread of agentic AI is creating an execution gap, with many organisations deploying tools faster than they can control them. TechRadar has reported that Gartner expects Fortune 500 companies to have more than 150,000 AI agents in production by 2028, while only a small minority of organisations feel ready to manage them properly. The concern is that without strong guardrails, AI can create shadow systems, expose data and weaken audit trails.
That warning matters in procurement, where decisions often touch spending authority, supplier risk and regulatory compliance. Companies adopting agentic ERP therefore need clear human-in-the-loop controls, defining which actions agents may take independently and which must be approved. They also need clean data structures, because fragmented supplier records and inconsistent category hierarchies can produce poor recommendations instead of better ones.
Gartner’s latest research also suggests the skills challenge is mounting. According to the firm, demand for AI expertise in supply chain roles rose sharply between early 2023 and early 2026, adding pressure to already stretched hiring pipelines. That makes internal training and careful deployment even more important if organisations want to scale these systems responsibly.
For now, the strongest use cases appear to be the most repetitive and time-sensitive ones: supplier email handling, invoice exception management and high-volume sourcing workflows. Those are the places where agentic systems can deliver measurable gains without pretending to replace human judgement altogether.
Source: Noah Wire Services
