AD FIELD NOTESAI advertising · the work behind the output

Case files / generative film

Svedka: translate a human dance into a controlled AI performance

For its Super Bowl debut, Svedka and Silverside revived the Fembot with a new BroBot sidekick, using human movement as the base for a primarily generative-AI film rather than treating generation as an unsupervised shortcut.

Campaign period: 8 February 2026Source published: 5 February 2026Archive entry prepared: 15 September 2026

Retrospective coverage, prepared in September 2026 from the original record.

Svedka Fembot and BroBot campaign characters in a neon-lit Super Bowl visual.
Svedka’s AI-assisted Fembot and BroBot world, shown in Silverside’s project case. Svedka / Sazerac / Silverside AI; image published by Silverside.

Reported execution: motion stayed human at the root

Silverside describes the Super Bowl spot as a human-led, AI-powered production. Real dancers supplied the movement reference; iterative passes translated that choreography into Svedka’s robot characters while the team refined continuity, character consistency and finish.

The campaign gave the Fembot a new counterpart, BroBot, and used their dance as the route back to a familiar mascot. The production partner frames the work as primarily generative AI, not a conventional live-action shoot with a small synthetic garnish.

Our reading: controlled transformation is more legible than full substitution

The useful design decision is to preserve a visible source motion. Viewers can understand what the system is transforming, while the brand and production team can judge whether the output still carries the timing and attitude of the original performance.

That creates a stronger human role than a generic ‘AI made the commercial’ headline. The dancers contribute the expressive substrate; the model expands the character world; editors and finishers protect the final piece.

Production lesson: build a pipeline around the failure modes

Character-led AI film tends to fail at continuity, hands, faces, contact and motion. A custom pipeline with modular builds and multiple refinement passes treats those problems as production gates instead of hoping a single prompt will solve them.

The practical brief should name the reference performance, the approved character states, the scenes that need frame-by-frame review and the finishing work that remains human. That turns a novel workflow into something a production team can actually schedule.

Evidence boundary: a production case is not a reception study

Silverside and trade coverage establish the Super Bowl placement and describe the AI-assisted method. The cited sources do not provide an independent study of brand lift, sales or audience sentiment.

The record supports a process lesson about motion transfer and character consistency. It does not support the broader conclusion that primarily generated spots are automatically cheaper, faster or more effective than other forms of production.

Sources & limits

Silverside’s production case and AdNews coverage describe the workflow; campaign reception and any performance claims are not independently audited in the cited record.

  1. Silverside AI — Svedka Super Bowl — source publication date not stated.
  2. AdNews — Svedka’s AI-generated Super Bowl ad — 5 February 2026.
  3. Associated Press — Super Bowl commercials — source publication date not stated.

Send a correction with the passage and supporting source.

The weekly field note

One useful note for the next brief.

Campaigns worth studying, the work behind them, and questions to bring into your next creative review. Delivered weekly.

How we handle your email