E-proc is a big bang. AI agents are the opposite.
Why e-proc implementations fail and how AI agents offer a faster, lower-risk path for procurement.
I've seen an Ariba project take 18 months, cost millions, and end with half the company still buying through email.
Most e-proc projects fail or deliver less than promised. They require board-level buy-in, alignment across procurement, IT, finance, and every requesting department, and during all that time the result is zero. When it finally goes live, the system becomes a glorified form that nobody wants to use. E-proc is a big bang, and that's why adoption is so weak.
AI agents work on the opposite logic.
A negotiation agent, for example, can run on top of Ariba or directly on SAP in a matter of weeks. No need to replace systems or run an IT project, and the requester stays on the same workflow.
That changes everything! Instead of requesting budget for an 18-month project, you show what already works and expand from there. It's the opposite of a big bang: solve a specific problem, prove value, build credibility, and grow.
If you're still putting together a business case for a monolithic e-proc project, you should stop and ask yourself: what can I solve in 30 days, without depending on IT, without asking for buy-in from 5 directors, showing results before requesting more budget?