LiBiAI - CaseStudy

Product goal

Overview

As the Lead UX Product Designer at a GenAI Stealth Startup, I spearheaded the design of LittleBirdAI, an intelligent personal assistant that acts as a second brain—helping users surface important insights, manage tasks, and stay ahead of critical information.

The Challenge

Making AI Proactive, Not Just Reactive

While Little Bird was great at retrieving contextual information when asked, it lacked a proactive system to push relevant insights to users. The goal was to design a feature that:

Anticipates

user needs before they ask

Surfaces

critical alerts in real time

Feels

like a natural extension of thought, not a rigid tool

This led to the development of Little Bird Alerts, an AI-powered system that monitors key topics, behaviors, and user-defined triggers, bridging the gap between passive assistance and proactive intelligence.

Research & Insights

To understand how users wanted AI to enhance their workflow, we conducted user interviews and competitive analysis. Key takeaways:

AI as a Personal Companion

Users who integrated AI deeply into their lives treated it like a real person—using it for everything from work to personal relationships. (Example: A user had AI analyze text messages and even provide outfit recommendations.)

Decision Fatigue is Real

Many users wanted AI to handle small cognitive burdens (e.g., tracking relationships, reminding them of key conversations, or monitoring industry trends).

Current Tools Are Too Passive

Users expected AI to go beyond summarizing what they already know—they wanted it to proactively search, analyze, and alert them about new, relevant information.

Fear of Overload

While AI-powered alerts were appealing, users didn’t want to be overwhelmed with irrelevant notifications—precision was key.

Core Concept

The Proactive AI Assistant

Little Bird Alerts is a smart notification system that allows users to define custom AI triggers for monitoring personal and professional topics.

Intelligent Monitoring

Users can tell Little Bird what to watch for (e.g., "Monitor my relationship with Ocean," "Alert me when there’s new payroll info"). The AI continuously scans conversations, documents, and web sources to proactively surface insights.

Context-Aware Nudges

The system identifies gaps or missing information (e.g., "You haven't followed up on this deal in two weeks"). Subtle nudges encourage users to stay ahead of their workflow.

Hyper-Personalized Alerts

Users train their AI agent by interacting with alerts—refining what’s relevant over time. AI understands behavioral patterns (e.g., if a user often forgets to reply to emails, it can remind them).

Seamless Integration

Works across platforms (chat, email, browser extensions). Leverages existing natural language inputs—no need for complex settings.

Implementation Strategy

UX + ML Collaboration

To ensure the AI felt intelligent and human-like, I worked closely with ML developers to refine:  1. Context tracking – AI should connect the dots across different inputs.  2. Signal vs. Noise filtering – Prioritize high-impact alerts to avoid overload.  3. Personalization loops – Users should train the AI through simple interactions.

Working closely with the ML team

Iterative Design & Testing

Early User testing

Early users tested different alerting mechanisms (passive vs. push-based, mobile vs desktop, platform context etc).

Refined notification sensitivity

Refined notification sensitivity to avoid unnecessary distractions by collecting context from screens and repeated user behavior.

Conclusion

Little Bird Alerts wasn’t just about productivity—it was about intelligence. By making AI proactive, intuitive, and deeply personalized, we bridged the gap between tool and thought partner. Impacting the following:

Increased AI engagement

Improved

information recall where users felt more in control of their workflows, reducing missed follow-ups and lost insights.

Validated

scalability, designed the feature with a modular system to allow future expansion (e.g., plugin integrations, advanced sentiment tracking).