DataFrontier is a research lab building the infrastructure for Unified Personal Intelligence. We work at the intersection of large language models, structured data systems, and enterprise revenue operations — publishing what we learn, productizing what compounds.

Personal context is a primitive, not a feature. A note on the substrate that should sit between people and their agents — the primitives that compose it, and the open questions that decide whether it is real.

What outbound sales teaches the field about production agent systems: orchestration under real constraints, buyer-intent modeling, and the handoff from agent to human.

Early thinking on the data layer a personal-intelligence substrate would need: what to store, where it lives, and who holds the keys.