Mobile Dev Memo Podcast Season 7, Episode 11: Understanding Unity's Vector (with Felix The)
Mar 24, 2026
Felix The, SVP of Product and Engineering at Unity who leads ads, ML and runtime data efforts, explains Vector as Unity's ML-powered growth platform. He discusses integrating engine runtime signals for better predictions. They cover unified large models, generative creative exploration, personalized micro-cohort creatives, and using playables to de-risk soft launches.
AI Snips
Chapters
Transcript
Episode notes
Vector Rebuild Unifies Gameplay And Ads
- Vector is Unity's rebuilt AI-powered growth and UA platform that unifies gameplay, monetization, and campaign data into a single continuous-learning model.
- The architecture is optimized for ingesting large volumes of game runtime signals, enabling faster learning and better predictions for mobile game advertisers.
Runtime Signals Beat SDKs For Sequence And Quality
- Unity's runtime is the earliest instantiation in a game, so runtime signals provide higher coverage and cleaner event sequencing than third‑party SDKs.
- Cleaner, ordered time series avoids join-errors and enables causal and sequential models that rely on event order.
Long Tail Purchases Drive Value And Need Better Signals
- Modeling needs to capture long‑tail monetization because whales and rare transactions drive disproportionate revenue.
- With extreme class imbalance, simple classifiers that predict 'no purchase' look accurate yet fail to find the valuable minority.