A 106-second, eight-chapter science animation about the Red Army’s Long March was completed autonomously by AI in WorkBuddy, taking roughly 20–30 minutes and consuming fewer than 40 credits. We believe this matters more than “AI writing a script”: AI video tools are beginning to deliver projects that can continue to be modified, rather than a finished video that is difficult to edit.
What this is
This project did not ask a general-purpose model to write animation code from scratch. Instead, it connected “Weijie AI Map Animation” to WorkBuddy. Once MCP—a standard interface that lets AI assistants call external tools and data—was connected, the AI could read the API documentation on its own, look up place names, plan routes, create subtitles and background music, and then call the platform to render the final video.
Unlike placing a static map in CapCut or manually tracing lines with an After Effects plug-in, this approach pre-packages mapping, timeline, shot, and rendering capabilities. AI mainly handles decisions and orchestration. The output remains an editable project file: it can be adjusted through conversation or opened in an editor for detailed changes.
That said, the 106-second runtime and cost of fewer than 40 credits came from a single case and do not yet show stable costs for complex projects across different tools, subject matter, and rounds of revisions. Its claims of “zero errors and zero warnings” mainly apply to this server-side trial run and cannot be treated as proof of comprehensive accuracy in historical or geographical content.
Industry view
The important shift is that competition in AI video is moving from generating a few seconds of footage toward delivering a complete production workflow. When routes, subtitles, music, review, and export are packaged as standard interfaces, enterprises can involve existing AI assistants in video production without first training a dedicated model. This could lower the barriers to producing historical education, corporate training, and marketing videos.
The risks are equally clear: mapping tools can improve execution efficiency, but they cannot automatically eliminate factual errors or aesthetic bias. If place names or routes are wrong, the finished video may be even more misleading. As more platform examples emerge, AI may also increasingly reproduce fixed narratives and visual templates. In addition, the claim that MCP can connect to multiple assistants depends on continued compatibility between both sides. Toolchain lock-in and credit-based pricing remain commercial variables.
We do not interpret “no longer needing CapCut and After Effects” as proof that professional production has been replaced. Sophisticated brand films, detailed packaging, and frequent revisions still require professional teams. The real value lies in compressing standardized tasks that used to take considerable time, allowing people to focus on topic selection, fact-checking, and creative judgment.
Impact on regular people
For enterprise IT: Businesses can package mapping, subtitles, music, and rendering as internal capabilities for AI assistants to call. Even so, videos intended for public distribution still require workflows for fact-checking, copyright clearance, and publication review.
For individual professionals: Beyond writing prompts, operations, sales, and training professionals should learn to review AI projects, including checking locations, data, subtitles, and brand guidelines rather than merely editing text.
For the consumer market: These tools may increase the supply of map-based educational and marketing videos, making animated maps easier for viewers to encounter. At the same time, greater content volume will make it harder to identify template-driven production and hidden errors.