Answers are grounded in this paper and cite their sources.
Procurement and supply chains account for most organisational greenhouse gas emissions, yet these Scope 3 emissions remain the hardest to measure and reduce. Agentic artificial intelligence (AI), meaning systems that can act with limited human intervention, is increasingly promoted as a solution, but it is unclear whether the evidence supports entrusting it with procurement decisions that carry financial, legal and carbon consequences. This review asks what research on AI-driven decarbonisation shows and what it implies for Net Zero procurement. Following PRISMA 2020, we reviewed 167 Scopus-indexed studies (2020–2026), combining topic modelling with qualitative synthesis. Three research streams emerge: carbon data and measurement, autonomous decision-making, and energy-system optimisation. AI has progressed from analysing data, to monitoring systems in real time, to acting autonomously, but almost entirely in buildings, electricity systems and transport; studies of autonomous procurement decisions are effectively absent. We interpret this as a delegation readiness gap: AI capabilities have advanced faster than the conditions needed to delegate procurement decisions responsibly. Three conditions define this gap: decision-grade supplier emissions data, clear decision authority with human oversight, and carbon accounting that includes AI’s own computational footprint. An integrative framework sets out how these conditions could be met. For managers, the priority is to invest in data and governance before autonomy; for researchers and policymakers, the review defines what must be demonstrated before agentic AI can be considered credible for Net Zero procurement. The study shifts attention from whether AI can optimise to whether organisations can delegate.
No takes yet. Share an insight, caveat, or question.
Khan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: