Building an AI Presenter
Uphold needed a presenter for its Flexible Staking video series. Not a stock model. Not a real spokesperson. A fully original AI presenter, built from scratch. The brief was simple to say and hard to execute: create someone who could explain a financial product with warmth, clarity and credibility. She had to feel real before she said a word.
For a crypto product, trust matters fast. The presenter needed to feel approachable, but still authoritative. Warm, but not soft. Polished, but not artificial. Someone you could believe in a product interview. Someone who could make Flexible Staking feel simple.
That meant focusing on the details that make a scene believable. A lapel microphone clipped to her blouse. Natural skin texture. Real fabric. A seated interview posture. Soft interior lighting. A room that felt considered, not generated.
This was not about creating an avatar. It was about building a believable AI presenter system around a character people could trust.
The Process
We treated the AI presenter like a real casting and production decision. The work began with character development. Expression, posture, styling, lighting and environment all had to point in the same direction: credibility.
- Build the character: We created a photorealistic AI presenter who could stay consistent across stills, video frames and editorial formats.
- Make her believable: We focused on the small cues you would expect from a real interview setup. A lapel mic. Natural clothing folds. Relaxed hands. Warm light. Human imperfections.
- Design the setting: The final scene used a warm interview environment with a cognac chair, soft lamp light and a calm interior backdrop. Premium, but not cold.
- Move into video: Once the character and setting were approved, we created the final video assets using AI video generation, synthetic voice and precise lip-sync workflows.
The Result
The project turned a complex AI workflow into a usable campaign asset.
The final workflow brings together character generation, synthetic voice, lip-sync, video generation and editorial control. Each part supports the same outcome: a presenter who feels credible enough to explain the product, and flexible enough to scale.
For a financial product, that balance matters. Trust first. Technology underneath.
