Strategy Simplified S24E24: 80% of AI Projects Fail. These 2 Firms Explain Why.
Sep 25, 2026
Ugo Philippart, an AI and data consulting leader at Emerton Data, and Damien Chan, an Adastra executive focused on enterprise transformation, unpack why most AI pilots never reach production. They explore governance, data readiness, ROI, adoption, trust, performance measurement, ownership models, and how AI is reshaping consulting careers and entire business functions.
AI Snips
Chapters
Transcript
Episode notes
Why AI Pilots Die At Production
- Pilots commonly stall because governance, MLOps, security, incident management, and data pipelines cannot support production.
- Legacy APIs and mainframes can take 12–18 months to modernize, while business leaders expect AI results much sooner.
AI Failures Have Multiple Root Causes
- Failed pilots usually reflect misaligned economics, weak foundations, poor adoption, or underperforming technical design rather than one universal defect.
- Business-led systems can be built on sand, while IT-led systems can fail because users and operating complexity were ignored.
Salvaging Failed AI Requires Rebuilding The Core
- Ugo Philippart’s team often salvages failed systems, but sometimes replaces 90% of the work while retaining only valid prerequisites.
- Fixes frequently require redesigning the model, tooling, and pipelines, even when user training and intended workflows remain useful.