Storytelling can reduce resistance to health messages by increasing identification, but the narrative elements, persuasive strategies, and emotional features that make recovery stories effective are poorly understood, thereby limiting real-time story generation for behaviour change. To address this gap, we developed and evaluated an artificial intelligence (AI) agent for generating persuasive recovery stories in real-time. From 156,871 Reddit posts in a depression community, we identified 988 inspiring recovery stories and used an explainable large language model to extract their narrative elements, persuasive strategies, and emotional trajectories. We encoded these elements in a retrieval-augmented generation (RAG) agent and evaluated 100 generated stories with mental health experts. Ratings showed near-perfect agreement (Cohen’s κ = 0.993, p < .001) and supported the stories’ clinical relevance and potential to motivate positive change. The next step is to integrate our AI agent into persuasive technology (PT) to enable real-time generation of tailored and clinically grounded recovery narratives that motivate behaviour change at scale, with potential applicability beyond depression.