BRISTOL MYERS SQUIBB
At a high level, the transformation has meant consolidating procurement processes and data onto a simplified technology stack, putting automation into the transaction layer and giving procurement teams real-time visibility into spend and supplier risk instead of fragmented, market-by-market views. In 2025, a unified, AI-infused procurement interaction layer was added.
“ The platform applies AI at the points where procurement historically relied on manual judgement calls,” she says, citing invoice matching, exception handling and supplier risk flagging as examples. Those decisions now happen automatically when the underlying data is clean and clear-cut, with cases routed to a person only when genuine judgement is required.
Touchless invoicing
Scott Hofmann and faster approvals
Chief Revenue Officer
Through these efforts, BMS has
GFT achieved 70 % touchless invoice processing through end-to-end digital workflows, alongside a 50 % reduction in purchasing approvals. Sharmita says these figures are progress rather than an endpoint. The improvements, she explains, stem from treating data quality as the foundation rather than an afterthought:“ You can’ t automate matching and approvals reliably on inconsistent supplier or PO data.”
BMS standardised its underlying data model first, then layered automation and AI on top of it market by market rather than attempting a single global cutover. The reduction in approval time came as a byproduct of removing friction at points where humans had previously been needed only because the data was not reliable enough to trust automatically. bms. com 9