This research investigates how working with large language models improves the regenerative capabilities of supply chains by developing digital process transformation capability and cognitive supply chain capability, under varying levels of organisational digital experimentation culture and artificial intelligence (AI) governance maturity. This study develops and tests a multi-stage capability architecture, introducing new perspectives on about cognitive automation, AI capability settings, and algorithmic affordances. The model is validated through analysis of responses from 281 respondents in knowledge-intensive fields. Empirical research supports the proposed serial mediation, depicting that incorporating large language models in supply chains enhance regenerative capacity through digital process transformation and the reconfiguration of cognitive supply chains. Digital experimentation culture strengthens the relationship between the large language model integration into supply chain and digital process capability, whereas AI governance maturity strengthens the link between such integration and regenerative capability. This research adds to modern theories on algorithmic cognition and capability orchestration in AI-enabled systems, adds depth to digital operations and strategic management research, and demonstrates how large language model integration can create regenerative supply chains.