Automating the repetitive part of animation.

The repetitive work

Animators were rebuilding the same kind of sequence word after word.

Pronunciation and spelling activities needed animation components assembled into a precise order. The creative components already existed, but choosing, arranging, and timing them repeatedly consumed animator time.

The useful question was not whether all animation could be automated. It was whether the predictable assembly work could be removed so animators could spend more time on the parts that needed their craft.

The hard part

A letter could not be mapped to one animation in isolation.

The selected sound and animation depended on the letters around it and, in some cases, what had already appeared. Letter pairs could produce one sound, and the same character could require a different component in a different word. I modeled those decisions as a contextual, deterministic ruleset.

The tool

Choose a word and follow the rule path.

This simplified interaction shows the shape of the original problem. The production rules covered the content’s real spelling and pronunciation cases; this example keeps the same idea small enough to inspect.

The output was predictable and reviewable. Given the same word and rules, the tool chose the same components every time. This was intelligent automation before modern generative AI, and it stayed honest about what the software was doing.

Impact

The rules engine handled the repeated assembly and left creative work with the animators.

The exact value was not one dramatic output. It was avoiding the same manual decisions across a large body of content.

Why it matters

The project shows the same judgment I bring to current Applied AI work.

I had to identify which part of a human workflow was stable enough to automate, represent the decisions in code, make the output inspectable, and leave the work that required human judgment with the people who owned it. The technology has changed; that responsibility boundary has not.