Using AI to Turn Human Stories Into Insight (and Build Trust)

On the Whole Whale podcast, George interviews Andy Citizen, CTO of Share More Stories, about collecting 150–300+ word experience-based stories (typed or voice) from targeted constituencies via a web app and analyzing them with AI. Share More Stories uses sequential classification across ~70 cloud models to score stories for evidence of emotions like anxiety, joy, and self-transcendence, combining these scores with light demographics and survey data, then using a generative AI agent constrained to the dataset to explore themes, anomalies, and demographic differences iteratively. They discuss nonprofit uses such as before-and-after journaling and program impact, the importance of prompts, and how AI should augment rather than automate, with emphasis on user competency, intent, validation, and avoiding hallucinations. Citizen critiques “AI everywhere” features and AI-written social content as trust-eroding, argues trust is a major opportunity, shares concerns about DevOps at scale, and reflects on community involvement and moving faster by reducing process gaps. 00:00 Meet Andy Citizen 00:50 What Share More Stories Does 01:57 Collecting Real Experience Stories 02:41 AI Scoring Emotional Signals 04:14 How Stories Are Gathered 05:24 Prompts Versus Surveys 06:26 Dashboards Reports And Agents 09:20 Nonprofit Program Evaluation Uses 12:06 Prompt Craft And Hidden Insights 14:05 AI Adoption And Training Gaps 17:46 Use Cases Lanes And Hallucinations 20:27 AI Side Of Fries And Trust 25:44 Rapid Fire Tech Questions 29:39 Personal Advice And Community 31:58 Magic Wand For Human Connection 34:12 How To Connect And Wrap Up