AI
TLDR: Bristol Myers used AI to cut procurement timelines from months to weeks, banking on faster decisions despite debated data readiness. The change targets speed and cost control across its supply chain.
Key Takeaways:
- Bristol Myers treated procurement as a bottleneck where slower approvals and data gaps usually win.
- AI helped shrink procurement timelines from months to weeks while forcing teams to assess data readiness assumptions.
- Faster buying decisions can tighten supply risk control, but the real test is sustaining quality as data improves.
Procurement usually runs on paperwork inertia, so dropping from months to weeks is the point. The interesting bet is that speed can still come first, even when data is not perfectly prepared.
Procurement usually runs on paperwork inertia, so dropping from months to weeks is the point. The interesting bet is that speed can still come first, even when data is not perfectly prepared.
Q&A
What procurement step is most likely to break first when timelines compress, even with AI support?
Supplier onboarding and contract alignment, because legal and quality checks often do not scale as quickly as model driven sourcing decisions.
If AI needs cleaner data, why might teams still accept imperfect readiness to gain early wins?
They can run hybrid workflows where AI suggests actions and humans validate, using early deployments to expose data gaps and prioritize fixes.
How does cutting cycle time change leverage with vendors in pharma procurement?
Shorter approval windows can improve Bristol Myers negotiating timing, but it also raises pressure on vendors to meet faster response requirements.
What metric beyond cycle time would determine whether the overhaul improves outcomes instead of just speed?
Purchase order accuracy, supplier compliance rates, deviation rates in quality release, and cost variance versus target over multiple cycles.
Could this approach reshape how pharma companies think about AI implementation across other operational functions?
Yes, if Bristol Myers proves that incremental AI deployments with human checkpoints outperform waiting for perfect data across functions like demand planning and inventory control.
No comments yet. Be the first to share your thoughts!