Episode 7 of Agent Mode AI. Abby and Avery walk the four datasets that document the bimodal ROI distribution in enterprise agentic AI: AM-029 on the Stanford Digital Economy Lab 12/88, AM-132 on the restored bimodal framing, AM-128 on the MIT NANDA GenAI Divide 95% finding, and AM-053 on the McKinsey State of AI 2025 17% EBIT-attribution. Then the GAUGE framework — six dimensions that distinguish the 12% high-performing cohort from the 88% struggling body. Governance, audit substrate, use-case maturity, guardrails, evidence baseline, exit posture.

Sources cited:

- Stanford Digital Economy Lab Enterprise AI Playbook 2026 (Pereira, Graylin, Brynjolfsson)

- McKinsey State of AI 2025 (n=1,993, November 2025)

- MIT NANDA State of AI in Business 2025 (Project NANDA / MIT Connection Science, August 2025)

- Gartner Q1 2026 Infrastructure & Operations Survey

- Fortune coverage of MIT NANDA findings, August 2025

Claims tracked:

- AM-029 — Why 88% of agentic AI deployments fail — agentmodeai.com/holding/?claim=AM-029

- AM-132 — The bimodal ROI distribution in enterprise agentic AI — agentmodeai.com/holding/?claim=AM-132

- AM-128 — The MIT 95% GenAI-pilot-failure claim — agentmodeai.com/holding/?claim=AM-128

- AM-053 — The McKinsey 17% EBIT claim — agentmodeai.com/holding/?claim=AM-053

Newsletter and the full Holding-up ledger: agentmodeai.com