
Nov 18, 2026 · 34 min
A persistent AI layer could make context portable
The Missing Layer for Personal AI | First Person Labs’ Ken Liu
The episode examines whether a trusted personal intelligence can coordinate specialized agents without surrendering control of sensitive data.
- 1Second builds persistent context from digital activity, including preferences, relationships, priorities, and goals.
- 2A trusted personal layer could route context across specialized agents while keeping users in control of inference and storage.
- 3Privacy safeguards face a hard limit: reasoning-based systems remain vulnerable to prompt injection and costly continuous analysis.
Don't miss
Ken Liu describes a future in which one persistent personal intelligence coordinates across screens, audio, homes, and other devices.
The brief
Ken Liu of First Person Labs introduces Second, a persistent context layer designed to understand a person’s memories, preferences, relationships, priorities, and goals without constant re-explanation.
Second analyzes streams of digital activity—app use, typing, browsing, and connected services—to give specialized agents portable context and occasionally act before a user asks.
The central design problem is trust: Second must decide what to reveal while protecting sensitive information through local storage, encryption, confidential computing, and user-controlled inference.
Prompt injection exposes the limits of that protection, since a system reasoning about personal data may still be manipulated by instructions from other agents or people.
Liu imagines one persistent personal intelligence coordinating across screens, audio, homes, and devices, while First Person Labs currently narrows its product to texting and APIs.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.