Before/After Redesign: PersonalAI’s Dashboard — Simplifying Control over Personal Models

AI · 6 min read

Before the redesign, PersonalAI’s dashboard presented an overwhelming list of custom models, permissions, and usage logs in a single table. Users reported confusion about model precedence and which data was used for training. Support tickets about accidental sharing grew as the user base expanded.

Designers reorganized the interface into three focused surfaces: models (with onboarding state and clear precedence), privacy controls (with friendlier defaults and one-click revocation), and a lightweight insights panel explaining data usage. They introduced inline explainers and a ‘safe mode’ toggle that automatically constrained model access for new uploads.

After launching, the company saw a 27% drop in privacy-related tickets and a 14% lift in daily active users engaging with models. The redesign underlined the importance of sensible defaults and explanatory UI when users are managing powerful, personal AI artifacts.

PersonalAI also open-sourced its dashboard accessibility tests and a migration checklist, helping peers avoid common pitfalls when exposing control over ML artifacts to end users.