DataMagic: Authoring Data Videos through Declarative Multi-Agent Orchestration

The paper presents DataMagic, a system that authors data videos from raw tabular data through declarative multi-agent orchestration. It introduces DVSpec to unify charts, narration, and animations with data-bound references, along with a 'Generate-then-Orchestrate' multi-agent strategy for parallel scene generation and narrative optimization. Evaluations on 109 real-world samples show that GPT-5 achieves 2.13/5 with success rates between 48.62% and 86.24%, whereas DataMagic improves quality to 3.89 (+83%) with success rates above 95%. A user study shows a 79.7% reduction in task time.

Even the most advanced LLM (e.g., GPT-5) achieves only 2.13/5 with execution success rates between 48.62% and 86.24%.

DataMagic improves quality to 3.89 (+83%) with success rates above 95%, with the most significant gains in animation and narrative dimensions.

A user study shows that, compared to a conversational LLM workflow, DataMagic improves creation efficiency (79.7% reduction in task time).