Toma Itagaki

Implicit

2026-0228

AI’s malleability allows for near infinite personalization, but getting to true personalization is difficult. Explicit mechanisms have been explored through markdown files and memory layers, but the implicit personalization that happened with social media feeds has not been explored in the explosive age of AI.

This client-side n=1 personalization has yet to be explored, and will be achieved by one who controls a non-trivial amount of AI interactions.

Implicit aims to create a single-point-of-contact to reimagine the user-info-interaction to become a truly personalized access point to information. Through careful interaction design and feedback loops, the interactions will feel proactive yet respectful — like equipping AI with intelligent body language.

Core Technology

A powerful use case is reworking notifications. We manually arbitrate and evaluate notifications from independent sources (emails, messages, apps, feeds). However, post-AI interactions should promote two fundamental changes to notifications: personal-arbitration and smart timing.

  1. Personal-arbitration: AI can learn from our preferences. The same way social media feeds learned more about our attention than we can explicitly know, notifications can be arbitrated with better control and personalization.
  2. Smart-timing: the delivery of notifications is broken — everyone has do not disturb on. There is no reason why AI can revert this equilibrium to make digital channels lively and connected again. This requires learning implicit preference of interruptibility and receptivity from the user.

What This Looks Like

Initial prototypes:

  1. Backend of the smart-timing mechanism using multimodal sensor and activity data (GPS, motion, calendar, wearable)
  2. Audio and gesture interface for ambient smart-timing exploration. Turn-based interaction with dictation + TTS + audio-commands + motion-based gestures.

Ask

Looking for people excited about human-computer interaction, reimagining what AI will do to how we interact with information. Building for next-generation of people 25 and under.