Event Agenda Builder using AI

The Process

The project moved through five distinct phases. Each one built directly on the last - the audit informed the workshops, the workshops shaped ideation, ideation constrained the design, and two rounds of testing refined it.

Phase 01

Research & Audit

The project began with a clear OKR: simplify agenda building for event attendees. Before running any workshops or drawing any wireframes, I invested time in building a deep, evidence-based understanding of the problem space. This meant going through all existing research insights related to agenda building, analyzing user behavior data in MixPanel and Sigma, and conducting a thorough internal audit of the current agenda-building experience across all products.

This phase was less visible but foundational. Without it, the workshops that followed would have been speculative. With it, I could walk into the room with validated pain points, concrete data, and a clear picture of what was already known, and what wasn't.

Phase 02

Problem Discovery Workshop

Armed with that research, I planned and facilitated a problem discovery workshop with product, research, and design partners. The goal wasn't to brainstorm solutions; it was to collectively agree on the problem. Together, we defined the problem statement, surfaced the constraints we'd be designing within, identified who the users were and what we already knew about them, listed the open questions we'd need to answer, and made our assumptions explicit so we could track and test them.

The output was alignment - a shared, documented problem statement that the whole cross-functional team could point back to throughout the project. This step made every subsequent decision faster and less contentious, because we had already agreed on what we were solving for.

Phase 03

Ideation, Journey Mapping & Scoping

With a clear problem statement in hand, I moved into ideation. Using existing Jobs-to-be-Done research data, I created a detailed attendee journey map that deliberately extended well beyond the boundaries of Attendee Hub. The full journey I mapped included the pre-event registration experience within a separate platform where they could sign up for initial sessions, and then the in-event experience inside the Attendee Hub platform post-registration.

I facilitated a design ideation workshop with the design and research teams, using the journey map as a shared reference. Grounding participants in the attendee's full lifecycle, not just their in-app experience, led to a much richer set of ideas. The session generated concepts that spanned Attendee Hub and beyond, including suggestions for improving touchpoints in the registration flow. After the workshop, I wireframed the strongest concepts and presented them to design leadership and stakeholders, receiving positive and directional feedback.

But this created a new challenge - we had too many good ideas and no clear scope. To resolve this, I organized and facilitated an Opportunity Solution Tree (OST) workshop with my product partners and Design Manager. We stepped back to the OKR, mapped out the various opportunities around it, listed the solutions each opportunity could yield, and surfaced the assumptions behind each.

This exercise narrowed the project to its final scope - an AI-powered recommendation flow that personalizes session suggestions based on attendee interests, goals, and behavioral data. The OST workshop was arguably the most strategically important step of the project. It happened before a single high-fidelity screen was drawn, and it set the direction for everything that followed.

Phase 04

Design & Iteration

With clear direction, I began designing the solution. During the initial flow, we hypothesized that:

1. users would easily find the AI recommendation entry point, and
2. they would intuitively recognize session conflicts when reviewing recommendations.

The designs went through multiple rounds of iteration; reviewing with engineering to check feasibility, presenting to stakeholders for feedback, and refining until we felt confident enough to usability test the designs.

Phase 05

Usability Testing & Refinement

Working with our user researcher, we ran the first round of usability testing. It disproved both hypotheses.

Round-1: Test usability

Missed entry point

The AI sparkle icon on the floating action button was too small and unrecognized. Users missed the feature entirely on first viewing.

Fix: The FAB label now expands with readable text when the user lands on the schedule page, then collapses after a short delay, making the feature unmissable on first visit without permanently cluttering the UI.

Unclear content

The language in the goals selection step reduced user confidence and completion rates.

Fix: The copy was rewritten to be more direct and outcome-oriented, with clearer examples of what selecting each goal would mean for their recommendations.

Missed conflicts

Users expected an explicit warning when a recommended session conflicted with one already on their schedule. Without one, the majority didn't notice.

Fix: A high-visibility "Conflict" indicator was added to recommendation cards, plus a calendar view for reviewing and resolving overlaps before committing to a session switch.

Outcome & Reflections