Mobile Dev Memo Podcast Season 7, Episode 2: Using AI for audience insights and analysis (with Francesco D'Orazio)
5 snips
Jan 14, 2026 Francesco D'Orazio, President and Chief Product Officer at Pulsar, specializes in audience intelligence. In their discussion, he explores how AI revolutionizes audience insights, balancing existing user optimization with new audience expansion. Francesco highlights the importance of structured data for AI outputs, and how simulations can predict audience reactions effectively. He argues for integrating product and marketing insights, and advocates for a shift towards hypothesis-driven analysis to avoid testing paralysis. A fascinating peek into AI's future in audience understanding!
AI Snips
Chapters
Transcript
Episode notes
Audience Insights Predict Demand Not Just Conversions
- AI helps reveal group beliefs and momentum that predict behavior beyond simple "what happened" analytics.
- Francesco argues growth now depends on understanding demand formation, not just conversion mechanics.
Structure Inputs To Avoid LLM Semantic Collapse
- LLMs degrade when given tens of thousands of unstructured documents and produce low-fidelity analysis.
- Francesco recommends feeding LLMs structured context graphs of entities, relationships, and time to restore precision.
Feed LLMs Context Graphs Before Analysis
- Provide LLMs with context graphs: entities, relationships, time, and confidence levels before analysis.
- Use human researchers to craft these structured inputs so AI outputs stay precise and actionable.