A small team monitors approved purchase requisitions in SAP, then runs the entire sourcing process outside the system — finding vendors, reaching out by email, WhatsApp, and phone, collecting quotes in Excel, negotiating, and finally raising a PO back in SAP. The ERP holds the bookends; everything in between is manual and off-system.
Under manual load, the team sources fewer vendors and negotiates less hard than it should — weaker prices. On a ₹100 Cr+ spend, even small inefficiencies are large numbers.
This isn't "remove the humans." It's remove the grunt work and the leakage — so the team runs far more spend, far more consistently, with a full record of every step. Humans keep the decisions; the agent does the legwork and holds the line on process.
Picks up each approved requisition the moment it's ready to source.
Known suppliers plus newly discovered ones, contacted across email and WhatsApp — every time, without fatigue.
Responses captured into a structured comparison automatically — outliers flagged, no more scattered Excel tabs.
Runs negotiation rounds within set guardrails, pushing for better terms consistently — not only when someone has bandwidth.
Once a vendor is selected, it drafts the purchase order in SAP for human approval.
Every vendor contacted, every quote, every message, every approval — one auditable trail.
The agent reads approved PRs from SAP, runs sourcing-to-negotiation across vendor channels, and drafts POs back into SAP — every action permission-checked and logged on Gantry.
A high-value process running on Excel, email, and WhatsApp around an expensive ERP is everywhere. SAP was meant to be the system of record; an AI agent is what finally makes it one.
This is the AI Automation Platform thesis applied to procurement: digitise the operation that's living outside the system, then let the agent run it.
Slowly, expensively, and with no audit trail — in spreadsheets, inboxes, and chat. We don't advise on AI — we build the agent and run it with you.