High adoption, low impact, agents move to scale
McKinsey’s 2025 AI report paints a split-screen: widespread AI adoption, but limited bottom-line impact. Organizations are piling into agents and pilots, yet data quality, cloud architecture, and change management remain the choke points to scale. The strategy signal is clear: fix the plumbing and the processes before you add another shiny AI wrapper.
Points clés
- McKinsey’s 2025 AI report is summarized as a TLDR by How to Win in Predict on Medium, focusing on the practical gap between promise and delivery.
- 90% of companies say they use AI, but 67% remain stuck in pilot mode, underscoring the difficulty of moving from proofs of concept to production.
- Scaling AI requires good, clean, and accurate data, supported by a sound, primarily cloud-based technology architecture.
- 62% of organizations are experimenting with AI agents, while 23% are already scaling them, with adoption concentrated in tech and healthcare.
- Healthcare is flagged as a major opportunity for AI agents, reflecting high user curiosity and need in digital-first patient journeys.
- The impact gap is stark: 64% say AI helps innovation, but only 39% report real EBIT gains, highlighting weak translation from activity to outcomes.
- Real scale demands significant investment not only in technology but also in the tools, people, and processes wrapped around it; the report also signals traits of “high performers” (details paywalled).
À retenir
Want AI gains without the “all pilot, no profit” hangover? Start by cleaning your data like it’s spring and your CFO is coming over. Move your stack to a robust cloud architecture, budget for change management like it actually matters (because it does), tie projects to EBIT on day one, and only then unleash the agents—especially in healthcare if it’s your turf. Because nothing says “AI transformation” like 27 pilots and zero EBIT… and we’ve all seen that movie.
Sources
