CASE STUDY:
AI CREATIVE WORKFLOWS CUT PRODUCTION COSTS BY 80%
THE SCENARIO
Waze allows users to choose awesome custom voices and avatars for their turn-by-turn directions. It’s called “Pick a Sidekick”. It’s fantastic for user engagement, and no other navigation app runs a program like it. It’s one of Waze’s “secret sauces”, and leadership is committed to using the program to help scale users in global growth markets. But it comes with a problem…
THE CHALLENGE
“Pick a Sidekick” is expensive.
Waze needs to hire recording studios, engineers, actors, and also pay fees to re-license those actors’ performances after a set number of years. The model almost punished us for success. If our “Drive with a Cat or a Dog” campaign was a hit in the US (which it is… it’s our #1!), then scaling it into new languages/markets becomes exponentially more expensive and difficult. Worldwide scaling is both cost AND bandwidth prohibitive.
THE BRIEF
Build a creative AI workflow to:
slash costs
accelerate timelines
maintain sky-high quality standards in our new localized languages and accents: Italian, Dutch, German, UK and Australian English.
THE WORK
We found the ideal blend of humans and technology to solve the problem.
We hired five human vendor-translators to make sure we localized the navigation experience with 100% cultural nuance. You’d be surprised how differently cats and dogs are perceived in different cultures, and we wanted the the content to authentically shine in each hyperlocal environment.
We then deployed Google AI Studio as our creative hub to scale the operation.
First, we taught ourselves to use the tool. Then we trained our vendor-translators in it, allowing them to express their native language skills while creating fully AI audio output. Then we built rigorous QA controls to ensure our quality standards, building custom filters in Adobe Audition to ensure our audio met the exacting specifications required by our app.
THE RESULTS
We hit an RBI triple (if you’re American), or scored a hat trick (if you’re British). Three huge wins:
we cut the unit cost of a localized voice experience by 80%.
we halved the time it takes to make a voice, and doubled the capacity in our content pipeline.
we found a user base that was previously out-of-reach, at a quality level indistinguishable from human output.
They say you can’t have things “cheaper, faster, AND better”. In an AI world, that’s no longer true at Waze.
BUILDING ON THE EXTRAORDINARY POWER OF AI, WE ALSO MADE WAZE’S FIRST EVER GENERATIVE VOICE EXPERIENCE. IT’s AN AI CHARACTER CALLED “PROFESSOR FACTS”.
What if, every time your navigation app gave you directions, it also gave you fun piece of trivia? That’s what you get when you drive with “Professor Facts”. And with the power of A.i. voice-cloning, we can infinitely refresh the experience, so users never hear the same material twice.
We designed a suite of visual assets to let users find and select the experience.
AUDIO SAMPLE
Click below to hear some sample navigation prompts from Professor Facts, voiced by (ahem) Tom Cowell.
MOOD: “BRAINY”
Waze lets users select a custom icon to represent themselves in the app. We call them Moods. We made a “Brainy” mood for this experience.
Vehicle: MOBILE LIBRARY
When selecting the Professor Fact’s voice, Waze users can travel through the map as a 3D vehicle. We made a mobile library for the Professor. We thought it turned out pretty well.
CREDITS
Waze Creative: Tom Cowell / Jamie Fine
Waze Production: Gabi Mcdaniels
Visual Design: Burn & Broad
3D Modelling: Irresistible Studios
Waze Leadership: Ally Taylor, Tram Huynh, Jake Shaw, Tatyana Schneider, Tim Queenan
