Giving Pricify a recognisable face, voice and repeatable social-video format

Pricify needed to explain a new way of sourcing building materials in the few seconds social audiences give a video. The message had to feel clear and approachable to builders, tradespeople and DIYers—not like a conventional software demonstration or a generic piece of synthetic content. A recognisable presenter could make the service easier to understand, but only if that character felt consistent and credible.

Long-form generative video introduces its own production problems. Facial identity, clothing, studio details, camera position and lip synchronisation can drift from shot to shot. A single forty-second generation would offer too little control, while independently generated clips could easily feel as though they belonged to different films. The spoken wording also had to remain exactly as approved.

We developed Dave as a reusable digital spokesperson, with a defined face, voice, hi-vis wardrobe and workshop studio. The approved narration became the timing master and was divided into five natural sections: the problem, an introduction to Pricify, how the service works, the available options and the call to action. Each section was short enough to generate and assess independently without changing a word of the script.

The shot plan alternated between front-facing and three-quarter views to give the edit movement while preserving continuity. Prompts specified restrained gestures, natural blinking, accurate eyelines, subtle camera movement and precise lip, jaw and cheek synchronisation. Low-cost previews created an approval gate before the higher-quality renders, with every shot checked for identity, hands, expression, background coherence and audio timing.

Voxd built a reference-guided AI video workflow around fixed audio and repeatable visual controls. The same character references, studio references, seed and identity-retention settings carried across the five shots. Each approved clip was rendered at 24 frames per second, trimmed around its real speech boundary and joined with clean hard cuts so the alternating angles felt intentional.

The complete master preserves the original voice recordings and their natural pauses, while a final finishing stage provides both a faithful enlargement and an AI-enhanced comparison render. Prompts, seeds, source mappings, output settings and QA reports were captured in a manifest, turning the film from a one-off experiment into a documented production system that can support future Pricify content.

The finished piece is a 40.46-second vertical explainer in which one coherent presenter takes the viewer from the familiar frustration of contacting merchants to Pricify's one-message, multiple-supplier-quote proposition. Five individually controlled shots create visual variety without losing Dave's identity, voice or workshop setting.

Pricify now has more than a single film. It has a reusable character, voice reference, studio environment and proven shot-by-shot workflow for creating further social content with a consistent on-screen presence. The production method keeps approved language intact, makes quality review practical and allows a new film to build on established creative decisions instead of beginning again from scratch.