Northwestern Medicine Makes AI Prove Its Savings

CFO John Orsini says supply chain AI tools must show results or be turned off.

Updated on Nov. 11, 2026 in Jobs

Bold flat-color vector editorial illustration of stacked plain cardboard supply cases on steel shelving in a clean storeroom, conveying careful inventory and spending control.
Northwestern Medicine requires AI supply chain tools to prove savings or be turned off, and has frozen new AI applications while it checks results. AI Illustration. Upload story photo >

Live Poll

Do you trust your local hospitals to spend on AI only when it saves money?

Northwestern Medicine has set rules requiring AI supply chain tools to prove savings, and has turned off tools that underdelivered while freezing new applications. Executive Vice President and CFO John Orsini described the approach during a panel at Becker's 14th Annual CEO + CFO Roundtable.

Why it matters

Health systems are under pressure to control costs, and Northwestern's approach ties AI spending to measurable returns rather than adding to the cost structure. The stance reflects concern that software maintenance costs accumulate and vendors sell overlapping point solutions.

Stanford Health Care's 26 AI use cases over about two years saved 17,000 hours and $8 million to $10 million in hard-dollar savings. A nurse infection-data app that cost $1 million to build and $250,000 to $300,000 a year to run reduced nurse abstraction time from eight hours to two.

The players

Northwestern Medicine

A Chicago-based health system that is testing AI supply chain tools against measurable savings targets.

John Orsini

Executive vice president and chief financial officer of Northwestern Medicine, who set the proof-of-savings rules for AI tools.

Stanford Health Care

A Palo Alto, Calif., health system that implemented 26 AI use cases over about two years.

Providence

A health system headquartered in Renton, Wash., whose chief supply chain officer Amanda Chawla was also cited in the discussion.

The details

Northwestern Medicine tests AI tools by requiring proof of savings or the tools are turned off, and it worked down its matched exceptions before deploying AI. Its accounts payable portal lets vendors take a discount for faster payment, and its next frontier is pharmacy, where it plans to tighten inventory and rebate processes with AI. A safety-improving supply chain project is funded ahead of projects competing on return on investment.

Timeline

  1. Expected meaningful results versus projections are anticipated in year two after installation.

  2. Stanford Health Care implemented its 26 AI use cases over about two years.

Roadmap

Northwestern's proof-of-savings standard places AI supply chain spending under the same return-on-investment scrutiny applied to other hospital capital projects. The approach contrasts with adopting tools for their own sake, as health systems weigh accumulating software maintenance costs against promised efficiencies.

For health system staff, AI tools that fail to deliver savings may be shut off, which affects which workflows and tools are available day to day. Projects that improve safety are funded before projects competing solely on return on investment.

The takeaway

Health systems can test AI investments by setting clear savings targets before expanding them, rather than adding tools that raise costs without proven returns. Tracking results against projections from the start makes it easier to decide which tools to keep.

Further reading

For more on hospital staffing and operations, browse Nursing Jobs in Chicago.

Source note: This article includes information reported by Becker's Hospital Review | Healthcare News & Analysis.

Live Poll

Do you trust your local hospitals to spend on AI only when it saves money?