The Agent Ledger
Real, measured AI agent outcomes.
A running record of what teams are actually getting from AI agents in regulated and high-stakes workflows: the numbers, by industry, with sources. No hype, no vaporware.
780,000 Cases, 42,500 Hours: What the DWP's AI Eligibility Pilot Actually Measured
The UK's DWP ran AI-assisted eligibility processing at real scale. The numbers are worth studying, including the ones that show where it broke.
What Walmart, Amazon, and Target Actually Measured When They Rebuilt Demand Forecasting
Three named retailers, three sets of hard numbers. What AI-driven demand forecasting actually moved, and what it means for inventory governance.
What AI Agents Are Actually Doing in Hospital Revenue Cycle
HCA saved $21M in error rework. Banner freed 30+ hours a week. Here's what governed automation looks like in RCM.
Grid Inspection by the Numbers: What Dominion, FirstEnergy, and ESO Actually Measured
Three utilities ran AI-assisted grid inspection programs and published real numbers. Here's what they found and what it means for asset management.
What Customer Service AI Actually Delivers: Five Companies, Real Numbers
Bank of America, NIB, Banco Itaú, Liberty, and Fairmoney all have published numbers. Here's what the results actually show.
Drug Discovery Is Running a New Clock
Sumitomo hit Phase I in 12 months. Insilico spent under $2.6M to reach Phase II. The timelines are real. Here's what the numbers actually show.
What AI Agents Are Actually Delivering in Warehouses
Amazon, DHL, and Walmart have posted hard numbers on AI-driven logistics. Here's what the results actually look like.
Sepsis Prediction AI: Three Health Systems, Three Measured Outcomes
UC San Diego, Johns Hopkins, and Duke each cut sepsis mortality with AI early-warning systems. Here are the actual numbers.
What Predictive Maintenance AI Actually Delivers: Numbers from GM, Siemens, and Unilever
GM cut unplanned downtime 60%. Unilever dropped maintenance costs 45%. Here's what the numbers actually look like in production.
15,791 Hours Back: What Ambient AI Scribes Are Actually Delivering in Hospitals
Kaiser, Cleveland Clinic, and others have published real numbers on ambient AI scribes. Here's what the data shows, and what it means for governance.
What 60% Faster Contract Review Actually Looks Like in Practice
Two real law firm outcomes on AI-assisted contract review: what the numbers mean, and what they demand from your governance setup.
What AI Agents Are Actually Doing in Insurance Underwriting
Named outcomes from Zurich and Lemonade, plus what industry benchmarks reveal about where AI agents earn their keep in underwriting.
Credit Union AI in Lending: Three Numbers Worth Tracking
FORUM, Centris, and Community Service CU logged measurable AI results in lending and member acquisition. Here's what the numbers actually show.
From Three Weeks to Minutes: What Claims Automation Is Actually Delivering
Real numbers from Lemonade, Frende, and a 400,000-claim travel insurer show what claims automation looks like when it works.
AML's False Positive Problem: What HSBC, Barclays, and Banco do Brasil Actually Measured
Three named banks, three audited outcomes. Here is what AI-driven AML monitoring delivered when measured against legacy rule-based systems.
What AI Fraud Detection Actually Delivers: Three Numbers Worth Knowing
PSCU, American Express, and PayPal have published real fraud-detection results. Here's what the numbers say, and what they demand of your governance model.