Strategy Summit 2026: Why AI Transformation Needs a Human Touch

What It Covers

Publicis Sapient CEO Nigel Vaz, speaking at HBR's Strategy Summit 2026, argues that AI functions as a business operating system rather than a technology tool, requiring organizations to redesign decision-making speed, business models, and cross-functional data flows before competitors do.

Key Questions Answered

• •AI as Operating System: Treat AI the way organizations treated the Internet in the 1990s — not as a technology upgrade but as a full redesign of how value is created and delivered. This means revisiting annual strategy cycles, multi-year planning horizons, and annualized budgets, all of which are too slow for AI-era competition.
• •Proof-of-Concept Scaling: Most AI pilots fail to scale because they solve narrow functional problems rather than representative organizational ones. Select problems large enough to signal broader transformation potential but small enough to deliver value quickly — the automotive example of compressing an 18-month car redesign cycle down to 18 weeks illustrates this sweet spot.
• •Anti-Linear Strategy Design: The single most common strategy killer is sequential, siloed thinking — corporate strategy feeding finance feeding marketing feeding operations. AI value emerges from connecting data across historically separate functions, such as using sales and marketing datasets to solve manufacturing problems, requiring intentional cross-functional data architecture from the start.
• •Unit-Economics Measurement: Replace milestone-based project reviews with continuous micro-metrics. Track cost per software release, cycle time per feature, and defect escape rate rather than waiting for annual budget cycles to validate strategic hypotheses. These granular signals allow real-time inference about whether transformation is moving in the right direction.
• •Ethics Embedded in Technology: Responsible AI requires translating ethical principles directly into technical decisions — specifying which data leaves the organization, choosing between open-source and closed models based on training transparency, and enforcing data sovereignty rules. Employees using public AI chatbots with customer data because internal tools are inadequate represents a preventable, still-common failure mode.