Aphra

Aphra

Aphra is a proactive personal-AI product that turns signals across email, calendar, reminders, and connected services into prioritized, per-user insights and reviewable actions. As Lead Engineer and Backend Owner, I worked across the intelligence pipeline, integrations, action flows, and mobile and web delivery that made the assistant useful beyond a chat interface.

Challenges

• Combining signals from different services into a coherent user state without turning every notification into noise.
• Capturing, prioritizing, and queueing insights per user while accounting for duplicates, stale context, retries, and partial integration failures.
• Moving from reactive AI chat to a proactive loop that could identify time-sensitive items before the user knew what to ask.
• Making suggested email and calendar actions feel fast while preserving an explicit human decision boundary.

Solutions

• Built the backend flow around structured signals and per-user insight queues, separating ingestion, context, prioritization, and delivery.
• Connected email, calendar, and contextual sources through OAuth-backed integrations while keeping failure and recovery paths visible.
• Designed the action lifecycle as suggest, inspect or edit, then approve or dismiss - keeping consequential execution under user control.
• Shipped the same operating model across mobile, web, APIs, identity, and integrations so proactive assistance remained understandable in daily use.