Return on AI: Weekly Digest — September 14, 2026

Seven findings this week, spanning manufacturing, retail, financial services, and AI infrastructure. Midea used AI to compress air conditioner design cycles enough to reach European markets ahead of competitors, while convenience stores running AI-assisted SKU planning are seeing 2 to 5 percent margin improvements. On deployment and returns, McKinsey finds that only 37 percent of firms see any EBIT lift from AI despite broad agent adoption, and Forrester's data points to factory floors as the environment where agents deliver measurable value over office settings. Globant cut AI spend growth to 2.2x while growing token usage 6.7x by routing requests intelligently across models, and Westpac reduced a data migration that previously took six weeks to a matter of hours using AI agents.


Midea Used AI To Cut AC Design Cycles And Enter European Markets First
80% Of Fortune 500 Run Agentic AI But Few Scale It
Convenience Stores Gain 2–5% Margin Lift With AI-assisted SKU Planning
McKinsey Finds Only 37% Of Firms See AI Lift EBIT Despite Agents Spread
Globant Grew Token Use 6.7x While Spend Rose Just 2.2x Via Intelligent Routing
Forrester Finds AI Agents Deliver More Value On Factory Floors
Westpac Cut Data Migration From 6 Weeks To Hours Using AI Agents

Key Takeaway of the Week

Westpac compressed a six-week data migration to a matter of hours using AI agents, the clearest proof this week that operational speed, not headcount reduction, is where AI delivers its sharpest financial returns. This result aligns with Forrester's finding that agents produce measurable value in structured, high-volume workflows like factory floors and back-office operations, while McKinsey's data showing only 37% of firms lifting EBIT confirms that broad deployment without process fit produces little. CFOs should treat time-to-completion on data-intensive workflows as the primary metric for evaluating agent investments.


Sponsored by Jetson Collective