Jellyvision

Jellyvision

Making decision support software more helpful and fun

Jellyvision helps people make informed decisions about their health and finances, using a mix of behavioral science and entertainment to make these topics easier to understand. I produce the media and logic for our interactive conversations, which simulate a chat with a real person. But what does that entail?

One of the trickiest parts of developing interactive conversation is making it, well, conversational. Software has increasingly demanded rich visuals, sound, and interaction patterns to engage users, a challenge designers feel acutely when dealing with important-but-eye-glazing topics — like, say, your benefits.

Part of my role at Jellyvision is creating the illusion of a person talking to you, using proprietary software with an HTML5/JS runtime to blend audio, visuals, and user input together into a sort of interactive video. But this work is more than technical: every line of text needs to be clear and relevant, every second of audio needs to sound natural, and every interactive element needs to be worthy of a click or tap. When it all works, the results speak for themselves: millions of people have benefited from talking with a Jellyvision bot.

A recent example: One of Jellyvision’s customers, a Fortune 500 retailer, wanted to explain some important benefits they offered that weren’t well-known or easily discoverable. Online doctor visits, a nurse concierge, budgeting and lifestyle apps — all helpful services, but not relevant to every employee. So how do you make a useful, actionable recommendation?

I was part of the team that developed the retailer’s benefits ambassador: an interactive conversation that recommends health and lifestyle benefits based on an employee’s needs. Specifically, I built the recommendation calculator that maps personal questions to tailored results. The recommender works with plans across all 50 states, factoring in a gobful of state-specific plan configurations and program availability. The result: personalized recommendations, with clear and immediate signup links, for thousands of employees across the country.