BackPortfolio / AI mascot
The warehouse brought to mind grim men aged 40–60. We broke that stereotype with the AI characters Sasha and Lena — and the number of people willing to take a shift grew sixfold.
Two people a week were showing up for warehouse shifts — that number is where the challenge started. The brief called for reaching 20–25 year olds, mostly men and a few less women, for roles as loaders, receiving and checking parcels, and entering data into a tablet.
Behind the ambitious goal was a perception problem no one had tackled for years: the warehouse was associated with grim men aged 40–60 — dirty, worn out, without a smile. Together with the marketing team we set out to break that stereotype and show young, energetic, content workers, so the audience could see themselves. Not hard labour, but movement, a team, a healthy pace.
That's how the idea of a character who would become the face of the campaign came about — a mascot. It had to embody the idea and make people want to join in: to be not just a worker but part of a team, and to achieve more than they had now.
I started with minimalist 2D: a drawn style, simple shapes, a basic palette. I was working from the idea of simplicity and easy perception. The result turned out flat in the literal sense — the character stirred no emotion, you didn't want to identify with it. It became clear: what was missing was volume, depth, physical presence.
I went back to prompt engineering and started working through hypotheses systematically: switching models, experimenting with CFG, seed and samplers, testing low-poly and realistic 3D styles. Each iteration sharpened the look.
Gradually a male character took shape, which people inside the company immediately nicknamed Sasha. I refined the details by hand: adjusting the geometry, adding logos, adapting the colours to the brand book. Then I used a neural network to build a full 3D model.
The character adapts with a single sentence: describe a pose, a mood, the objects around it — and you get the shot you need without rebuilding from scratch. That gave flexibility and saved a huge amount of time at the stage of producing ad materials.
Lena grew out of Sasha — a female character with the same editing logic. One shared system made it possible to build her just as quickly, without losing the recognisable style.
The campaign went live: the characters rolled out into advertising — banners, infographics, promo mechanics. Sasha and Lena became an invitation to join the team, not just a picture in a warehouse.
After a while the numbers showed the result: six times as many people wanted to take a shift — from two a week to twelve. Repeat shifts rose from 14% to 37%. Prompt engineering and pairing neural networks with hand refinement delivered what standard approaches couldn't.
Here AI didn't just process data — it gave a human, modern and relatable face to a job like warehouse work. The character became not a worker but an invitation to join the team — and that worked better than any direct offer.
